Popular Rationalism | September 2, 2026

Science located no antivaccine program in the Interagency Autism Coordinating Committee’s new strategic plan, so Sara Reardon’s August 31 article supplied one by inference. “In autism advisory committee’s funding plan, critics see a veiled antivaccine focus” acknowledges that the plan never mentions vaccines, then asks readers to treat its attention to regression, environmental exposures, immune biology, and committee members’ affiliations as evidence of what the document supposedly conceals. Alison Singer of the Autism Science Foundation states the method plainly: “We could read between the lines.”

That is a revealing sentence because the lines themselves do not establish the headline. A claim that a federal plan contains a veiled agenda requires evidence of how the veil operates. One would expect to find a vaccine exposure category, a funded vaccine protocol, an agency deliverable, a causal presumption, a surveillance variable, an endpoint, or at least language broad enough to direct investigators toward vaccination as the favored explanation. Reardon instead identifies Robert F. Kennedy Jr.’s views, the backgrounds of public committee members, and the personal beliefs of some parents on the committee. Those facts can justify scrutiny of the committee, but they cannot do the evidentiary work of showing what the plan directs the federal government to study.

Subscribe now

The article then weakens its own insinuation. The researchers Reardon quotes agree that regression is real, poorly understood, and scientifically worth studying. Sally Ozonoff tells Science that prospective observation should replace dependence on parental memory, which is also what the IACC plan says before specifying how to do it. The paper Science links to support its account of timing says that developmental declines emerge around the first birthday and continue through the second year, yet Reardon tells readers that measurable declines occur primarily within the first 12 months and therefore before most routine vaccination. Her linked paper contradicts the first assertion, and the federal childhood immunization schedule contradicts the second, leaving the article in possession of the evidence needed to test its framing but unwilling to apply it.

Share

The operative text points elsewhere

The official record requires one preliminary distinction. As of September 1, the IACC’s August 27 meeting page contains a 336-page August 17 updated working draft, proposed revisions, federal-member correction requests, and amendment documents. It does not yet provide a clean, consolidated post-vote edition. Any serious analysis should therefore identify which version it reviewed and avoid pretending that a working draft, an amendment packet, and a final integrated plan are interchangeable. The analysis here uses the August 17 updated working draft and notes the separate revision record where relevant.

A full-text search of that draft returns no occurrence of vaccine, vaccination, immunization, antigen, adjuvant, thimerosal, mercury, or aluminum. Absence of a word cannot prove the absence of every private intention, but a public funding plan is judged by its operative language. The relevant question is what the plan authorizes, recommends, measures, and funds. On those points the alleged vaccine focus never appears.

The plan instead adopts a heterogeneous model of autism. It states that autism arises through “multiple interacting developmental pathways” involving varying combinations of genetic susceptibility, biological vulnerability, environmental influence, and developmental timing. It warns against a single cause and says regression itself should not be treated as a single subtype or mechanism. Its environmental priorities are named: air pollution, pesticides, metals, endocrine-active chemicals, water contaminants, maternal health, placental biology, infection, medication, diet, nutrition, acute inflammatory or metabolic stressors, and cumulative mixtures. A reporter who believes vaccination is encoded somewhere in that list must identify the textual hook and explain why the listed exposures should be read as decoys, a showing Reardon never makes.

Subscribe now

The budget is equally resistant to the headline. The draft’s FY2028 table recommends a total federal autism portfolio of $747.444 million. The dedicated National Neurodevelopmental Regression Initiative receives $57 million, with another $10 million regression domain inside the broader Part IV portfolio. Even on the draft’s maximal budget presentation, the two explicit regression lines total $67 million, just under nine percent of the recommended portfolio. The remaining plan covers neural circuits, immune and inflammatory biology, folate metabolism, gastrointestinal and microbiome biology, mitochondrial and metabolic function, autonomic dysfunction, sleep, epilepsy, motor planning, adaptive trials, diagnosis, communication, mental health, education, housing, caregivers, employment, aging, and transportation. Calling this a veiled vaccine funding plan requires the reader to disregard almost the entire allocation.

Reardon also invokes committee members who promote leucovorin, which invites readers to carry those members’ clinical views into the plan. The plan’s own language blocks that transfer on page 91: “This Plan does not recommend leucovorin as a treatment for autism.” It keeps FOLR1-related cerebral folate transport deficiency, other causes of cerebral folate deficiency, and an unconfirmed folate-receptor-alpha-autoantibody subgroup analytically separate. It calls for assay harmonization and an adequately powered multisite confirmatory trial before autism-wide clinical guidance or coverage. That distinction matches the FDA’s March 10, 2026 approval, which applies to patients with a confirmed FOLR1 variant, not to autism as a diagnosis. The section may be debated on its scientific merits, but describing committee members’ practices does not erase what the plan actually says.

Science’s own experts supply the rationale for studying regression

Reardon quotes UCLA psychologist Catherine Lord saying that regression is real, heartbreaking, and poorly understood, with little evidence that current interventions prevent or reverse it. She also reports that Ozonoff welcomes research attention to the subject while insisting on prospective follow-up from early life. Those statements are a rationale for the research initiative, not evidence against it. A clinically consequential event that affects a substantial minority of autistic children, lacks a standard definition, and has almost no established neurobiology is a defensible target for coordinated research.

The broader literature makes the case more concrete. A 2021 systematic review and meta-analysis by Christine Tan and colleagues included 97 studies, with 75 studies and 33,014 participants contributing to the meta-analysis. It estimated a pooled regression prevalence of 30 percent, although heterogeneity was extremely high, and found a weighted mean age of onset of 19.8 months. The authors called for a standardized definition and repeated measurement across early childhood. Those are almost exactly the first tasks in the IACC proposal: a common case definition, harmonized phenotyping, prospective longitudinal cohorts, serial biospecimens, and repeated assessment before, during, and after developmental change whenever possible.

The plan also incorporates the methodological warning Science presents as grounds for suspicion. Page 61 says studies “should not depend strictly on retrospective parental recall” and directs investigators to combine medical records, geospatial data, prospectively collected specimens, validated or recorded exposure measures, and prespecified causal-inference methods. Ozonoff’s study of parent recall found low agreement between reports made when children were two to three years old and reports made when they were about six. One quarter of families changed onset categories, most often because parents later failed to remember a regression they had reported earlier. That finding warns against retrospective timing estimates without making regression imaginary or justifying the dismissal of skill loss reported closer to the event; it supports prospective observation, which is precisely what the plan proposes.

The draft contains mechanistic hypotheses that deserve close review, including immune, mitochondrial, metabolic, purinergic, genetic, gastrointestinal, epileptiform, and environmental pathways. Some are mature enough for structured testing; others remain speculative. The proper scientific response is to classify each by evidence stage, define the subgroup, prespecify the endpoint, and stop failed hypotheses, a discipline the plan repeatedly states. Critics can still argue that particular claims outrun their citations or that some proposed allocations are poorly weighted, but they must examine the claims and citations. Treating the very existence of environmental and immune research as a coded vaccine statement avoids the scientific argument.

The sentence that debunks the article

The most consequential reporting error appears in Reardon’s treatment of developmental timing. She writes that measurable declines “primarily happen within the first 12 months of life” and that this is “earlier than most children receive routine vaccines,” although the first clause is not what her linked source says and the second is incompatible with the schedule.

The linked review by Sally Ozonoff and Ana-Maria Iosif summarizes prospective infant-sibling studies. It reports largely intact early development followed by declines and symptom emergence “around the first birthday and in the second year of life.” It describes groups that look similar at six months, diverge by 12 months, and continue to diverge afterward. A statistically detectable group difference at 12 months does not mean that the underlying declines occurred primarily before 12 months, much less that they were complete by then. The review’s own summary states that decline occurs during the first and second years of life. Tan and colleagues’ 19.8-month weighted mean provides an independent check against Reardon’s compressed chronology.

Share

The vaccination claim fails by ordinary calendar comparison. The CDC’s January 2026 framework retained vaccines for pertussis, tetanus, diphtheria, Hib, pneumococcal disease, and polio in the category recommended for all children. The agency’s posted age table places the first DTaP, Hib, pneumococcal, and polio doses at two months, followed by doses at four and six months. Whatever causal question one wishes to investigate, routine childhood vaccination plainly begins before the first birthday and long before the 19.8-month mean reported in the meta-analysis. Calendar arithmetic and source fidelity settle this narrow point without deciding any causal question about vaccines.

This is where the article debunks itself most cleanly. It links a paper that places decline across the first and second years, converts the emergence of group differences by 12 months into a claim that decline primarily occurs within 12 months, and then places routine vaccination later than a schedule that starts in infancy. The resulting sentence is not a cautious summary of unsettled science. It is a timeframe substitution followed by a factual error about vaccine timing.

The rollout’s critiques have their say, and Science still misstated it

The committee’s process evidently gives critics some material. They claim a A 336-page plan should not be placed on an abbreviated review track, that thousands of public comments should not reach voting members less than a day before deliberation, and that a member who requested disability accommodation should receive time and access sufficient to participate. They do like that drafters lacked a complete institutional record and formal handoff, did not have the unreleased 2024 strategic plan, and relied on public materials they could independently locate. They do not like that Appendix A says one portfolio classification used a language-model classifier applying fixed rules, with only approximate reproducibility and without careful human review of every project. They say these are material limits because they affect continuity, agency verification, cost assumptions, and the reliability of the portfolio categories used to justify priorities.

One has to wonder if other U.S. agency committees are held to the same standard, and more on point, where these critics were during the long tenure of “speak-no-harm” Advisory Committee on Immunization Practices (ACIP) where the word “risk” was, evidently, banned speech if associated with vaccines.

Still, after 26 years of the IACC predecessor committees existing and doing virtually nothing about environmental factors that have been associated with autism, one has to acknowledge that this IACC has a lot to make up for.

Given the criticisms, Science had a strong process story available but weakened it with another chronology error. Reardon says IACC released the July draft four days before its next public meeting. The Federal Register notice scheduled the meeting for July 31. Contemporary reporting says the draft was released on Monday, July 20, which made the interval eleven days. Four days was the original public-comment period, as Disability Scoop reported on July 23. The distinction does not rescue the committee’s process; four days for public comment on 336 pages was indefensible. It does show that Science converted a valid criticism into an inaccurate sentence.

The August process remained compressed: thirteen autism organizations asked the committee to delay the vote until members could review the public input, federal members abstained in large numbers, and Scott Robertson stated that he could not process the material on the schedule provided. Those facts support criticism of governance and accessibility. They do not prove that a $57 million multi-pathway regression initiative is a secret vaccine project.

But even that logic defiles the prima facie reality recognized by anyone who bothers to examine the history of studies proferred as definitive on the question: Bona fide, good-faith studies of the role of vaccines in autism etiology are long overdue.

The burden Science never met

Committee membership matters because people choose priorities, interpret evidence, and control agendas, and Kennedy’s history makes vigilance warranted. A member’s financial or organizational ties can create conflicts that should be disclosed and managed. These propositions establish a reason to inspect the plan closely without determining what inspection must find. Reardon’s article reverses that sequence by beginning with the people, treating regression and environmental research as suspicious because of those people, and then using the absence of vaccine language as evidence that the agenda is veiled. This makes the claim resistant to disproof: explicit vaccine language would prove the agenda, while no vaccine language proves concealment. A claim structured that way cannot be audited and can only be believed or rejected.

An auditable approach begins by identifying the operative language, the funded exposure, the protocol, the endpoint, the agency instruction, and the causal presumption. If a later NIH solicitation turns “environmental exposures” into a vaccine study, the solicitation can be read and criticized on its terms. But why shouldn’t they? If CDC adopts a regression definition built around vaccination, that definition will be evidence - of something. If an agency steers money outside the plan’s stated exposure categories, the award record will show it. The August working draft contains none of those acts.

But we hope NIH funds vaccine studies to the hilt.

Science promised readers a veiled antivaccine focus. The document shows a broad autism research and services plan with an explicit, minority allocation for regression; the plan’s methods answer the caution urged by the experts Science quotes; the linked regression review contradicts the article’s 12-month formulation; and the federal immunization schedule contradicts its vaccine-timing claim. The committee’s rushed process merits direct criticism, and parts of the draft warrant stringent scientific review, but the hidden vaccine case still requires evidence. Reardon’s article did not produce it because the sources she cited support the regression research question and defeat the sentence used to make that question look suspect. That is how Science debunked its own framing of the IACC plan.

Leave a comment

Source record

1. Sara Reardon, “In autism advisory committee’s funding plan, critics see a veiled antivaccine focus,” Science, August 31, 2026.

2. Interagency Autism Coordinating Committee, August 27, 2026 meeting materials.

3. Interagency Autism Coordinating Committee, Updated IACC Working Draft Strategic Plan 2026–2028, August 17, 2026, especially pp. 26–29, 56–61, 68–70, 90–92, 255–260.

4. Sally Ozonoff and Ana-Maria Iosif, “Changing Conceptualizations of Regression: What Prospective Studies Reveal About the Onset of Autism Spectrum Disorder,” Neuroscience & Biobehavioral Reviews 100 (2019): 296–304. doi:10.1016/j.neubiorev.2019.03.012.

5. Christine Tan et al., “Prevalence and Age of Onset of Regression in Children with Autism Spectrum Disorder: A Systematic Review and Meta-analytical Update,” Autism Research 14, no. 3 (2021): 582–598. doi:10.1002/aur.2463.

6. Sally Ozonoff et al., “Reliability of Parent Recall of ASD Symptom Onset and Timing,” Autism 22, no. 8 (2018): 891–896. doi:10.1177/1362361317710798.

7. Centers for Disease Control and Prevention, “CDC Acts on Presidential Memorandum to Update Childhood Immunization Schedule,” January 5, 2026; and “Recommended Vaccines for Young Children,” childhood schedule by age.

8. U.S. Department of Health and Human Services, “Office of the Secretary; Notice of Meeting,” Federal Register, July 7, 2026.

9. Michelle Diament, “After Uproar, Feds Back Off Rushed Effort To Approve New Autism Plan,” Disability Scoop, July 23, 2026.

10. Autistic Self Advocacy Network, “Community Calls on IACC to Review All Public Input Before Strategic Plan Vote,” August 26, 2026.

11. U.S. Food and Drug Administration, “FDA Approves First Treatment for Patients with Cerebral Folate Transport Deficiency,” March 10, 2026.

Ask Google whether methylene blue can be taken with Alka-Seltzer and the answers quickly become contradictory. One result warns of a potentially life-threatening interaction. Another finds no interaction at all. Both answers can appear authoritative because the question contains a hidden error: “Alka-Seltzer” is a brand name, not a drug.

Alka-Seltzer Original contains aspirin, citric acid, and sodium bicarbonate. Several Alka-Seltzer cold products contain acetaminophen, dextromethorphan, phenylephrine, and sometimes an antihistamine. These products share a logo and a marketing lineage, but they do not expose the body to the same chemicals. Asking whether methylene blue interacts with “Alka-Seltzer” is therefore like asking whether a drug interacts with “something made by Bayer.” The exposure has not been defined.

That distinction matters because methylene blue has real pharmacological activity. It is widely discussed as a mitochondrial aid, cognitive enhancer, antimicrobial compound, and longevity supplement. Whatever merit those proposed uses may eventually prove to have, methylene blue is not biologically inert blue coloring. It is a potent, reversible inhibitor of monoamine oxidase A, the enzyme commonly abbreviated MAO-A.

Researchers established this property directly in laboratory experiments. In a 2007 study, Ramsay, Dunford, and Gillman found that methylene blue inhibited MAO-A at concentrations low enough to have plausible clinical consequences. They specifically warned that the effect had implications for oral exposure because MAO-A helps regulate serotonin and other monoamines in the brain and gastrointestinal tract. The paper did not merely classify methylene blue by analogy to antidepressants. It measured the enzyme inhibition. The study was published in the British Journal of Pharmacology.

This is where the identity of the Alka-Seltzer product becomes decisive.

Original Alka-Seltzer presents a different problem

According to the current Drug Facts label, each Alka-Seltzer Original tablet contains 325 milligrams of aspirin, 1,000 milligrams of anhydrous citric acid, and 1,916 milligrams of sodium bicarbonate. The usual labeled dose is two tablets dissolved in water. That supplies 650 milligrams of aspirin and 1,134 milligrams of sodium. The complete label is available through DailyMed.

None of those ingredients creates the serotonin interaction that makes some methylene-blue combinations dangerous. Original Alka-Seltzer contains no dextromethorphan and no phenylephrine. Its relevant risks come from aspirin and sodium: gastrointestinal bleeding, aspirin sensitivity, interactions with anticoagulants or other anti-inflammatory drugs, and a sizable sodium load for people with hypertension, kidney disease, heart failure, or a sodium-restricted diet.

That does not make Original universally safe. It places its risks in the correct category. A warning about serotonin syndrome would be misplaced for this formulation because the ingredient that creates that concern is absent.

The cold remedies are pharmacologically different products

A current Alka-Seltzer Cold day formulation lists three active ingredients in each capsule: 325 milligrams of acetaminophen, 10 milligrams of dextromethorphan, and 5 milligrams of phenylephrine. The nighttime version adds doxylamine. The label tells consumers not to use the product while taking a prescription MAO inhibitor or for two weeks after stopping one. DailyMed lists the ingredients and MAOI warning in the 2026 product label.

Dextromethorphan is the cough suppressant found in scores of over-the-counter products. It also affects serotonergic signaling. When serotonin transmission is increased while its enzymatic breakdown is simultaneously inhibited, serotonin can accumulate to toxic levels.

The prescribing information for pharmaceutical methylene blue names dextromethorphan explicitly. It states that opioids and dextromethorphan may increase the risk of serotonin syndrome and advises against combining methylene blue with drugs that enhance serotonergic transmission. Some reported cases have been fatal. The warning appears in the current ProvayBlue prescribing information.

Phenylephrine adds a second concern. It constricts blood vessels through alpha-adrenergic activity. Combining sympathomimetic decongestants with MAO inhibition can produce an excessive cardiovascular response, including sharply increased blood pressure. This is one reason cold-remedy labels carry an MAOI prohibition even when the consumer has never heard of serotonin syndrome.

The question is therefore no longer “methylene blue plus Alka-Seltzer.” It is methylene blue plus dextromethorphan, with phenylephrine and other ingredients potentially along for the ride. That is a much more serious question, and the official labels answer it conservatively: avoid the combination.

Dose and route matter, but they do not erase the mechanism

Most documented cases of methylene-blue-associated serotonin toxicity involved intravenous administration during surgery or critical care. Doses were often around 1 milligram per kilogram or higher, and many affected patients were also taking serotonin-reuptake inhibitors. A review of published reports found 25 patients who had received methylene blue while taking a serotonin-reuptake inhibitor or clomipramine and then developed an acute confusional state compatible with serotonin toxicity. That review appeared in the European Journal of Anaesthesiology.

Those cases do not give us a reliable incidence estimate for someone taking a few drops of oral methylene blue. Intravenous dosing produces different peak concentrations and tissue distribution. A low oral dose and a hospital infusion are not equivalent exposures.

The opposite mistake is to treat “oral” as a synonym for “negligible.” An aqueous oral formulation of methylene blue has demonstrated substantial systemic bioavailability. Human pharmacokinetic research documented high absolute bioavailability after oral administration. A 2018 report also described serotonin toxicity after initiation of an oral urinary analgesic containing methylene blue in a patient already taking several serotonergic psychiatric drugs. The authors found 50 earlier cases in the literature, all involving parenteral methylene blue, making their case an important warning that oral administration does not reduce the risk to zero. The report was published in Psychosomatics.

What remains unknown is the absolute risk from a particular low-dose oral methylene-blue product combined with a standard dose of dextromethorphan. There is no large trial that assigns people to that combination and measures serotonin syndrome. There should not be one. The known mechanism, official labeling, clinical cases involving related combinations, and availability of alternative cold treatments already make intentional coadministration difficult to justify.

Uncertainty about frequency is not uncertainty about whether the mechanism exists.

The labels do not provide a simple home washout rule

The prescription methylene-blue injection label advises patients not to take serotonergic drugs within 72 hours after the last dose. Cold-remedy labels use a broader warning against taking the product during MAOI treatment or within two weeks after stopping the MAOI.

Those intervals come from different regulatory contexts. They should not be blended into a homemade rule for oral methylene-blue supplements. The appropriate interval may depend on the methylene-blue dose, formulation, route, duration of use, kidney function, and the identity and half-life of the other drug. Simply separating the two products by breakfast and bedtime does not resolve the uncertainty.

Someone using methylene blue who needs treatment for a cold should read the active-ingredient panel, not the large print on the front of the package. A product containing dextromethorphan or a systemic decongestant deserves specific review by a pharmacist or prescriber. A single-ingredient alternative may avoid the interaction altogether.

If the products have already been combined, the sensible response is observation and dose-specific advice rather than panic. No additional doses should be taken until the formulation has been identified. Agitation, confusion, heavy sweating, fever, tremor, muscle jerking, overactive reflexes, rigidity, diarrhea, a racing heartbeat, marked blood-pressure changes, or seizure require urgent assessment. In the United States, Poison Control can evaluate the actual products, amounts, times, and symptoms at 1-800-222-1222.

The larger lesson is exposure precision

This small medicine-cabinet problem illustrates a failure that appears throughout health research. We classify exposures too broadly, then argue over conclusions produced by the bad classification.

“Alka-Seltzer” is too broad. “Methylene blue” may also be too broad unless the dose, route, formulation, purity, frequency, and timing are known. A risk assessment built from brand names rather than active ingredients has already lost the causal thread.

Search engines and automated interaction checkers often make this worse. A checker may attach the warning to the entire brand family and frighten someone who took Original. Another may examine only aspirin and sodium bicarbonate, declare that no interaction was found, and falsely reassure someone holding a Cold & Flu package containing dextromethorphan. Neither answer is rational because neither begins by identifying the exposure.

Original Alka-Seltzer does not contain the ingredients responsible for the methylene-blue serotonin interaction, although it carries substantial aspirin and sodium risks of its own. Alka-Seltzer cold and cough products containing dextromethorphan, phenylephrine, or both should not be casually combined with methylene blue.

The brand name belongs to marketing. The active ingredients belong to pharmacology. When the two disagree, pharmacology wins.

Share

Leave a comment

August 31, 2026

HHS says hospitals and clinics turned distressed minors into recurring customers, secured payment through questionable coding, and expanded under federal and professional protection.

On August 13, 2026, the Department of Health and Human Services released Wolves in White Coats: How Doctors and Hospitals Pushed and Profited from the Fraud of ‘Gender Medicine.’ The 64-page commissioned report draws on insurance claims, CMS coding guidance, Justice Department investigations, hospital materials, whistleblower testimony, scientific literature, and interviews with patients and parents. Its conclusion is stated in the title: HHS says pediatric gender medicine became an industry that harmed children and rewarded the institutions delivering it. ^[1],[2]

The release was paired with federal referrals. Vice President JD Vance, acting as chairman of the White House Task Force to Eliminate Fraud, referred providers named in the report to the Department of Justice. Secretary Robert F. Kennedy Jr. sent the same claims cohorts to the HHS Office of Inspector General. Kennedy’s August 12 letter identifies providers whose claims exhibit ‘potentially anomalous billing patterns’ and asks OIG to investigate possible violations of federal law. ^[2],[3]

Subscribe now

HHS describes a national system driven by three forces. Hospitals found a long-duration revenue model. Advocates and providers used diagnostic codes that could secure insurance payment when a gender-related diagnosis might fail. Federal officials and professional societies then supplied the legal, financial, and institutional protection that allowed the model to spread. The report’s five chapters build that case in sequence. ^[1]

The report’s central charge

HHS says hospitals enabled a rapid medical shift despite scant evidence for long-term safety and efficacy. The report uses the term ‘sex-rejecting procedures’ for puberty blockers, cross-sex hormones, mastectomy, genital reconstruction, and related operations. It argues that these interventions placed physically healthy minors on a path of repeated endocrine visits, laboratory monitoring, prescriptions, mental-health encounters, surgeries, revisions, and adult follow-up. ^[1]

The report was commissioned by HHS and written by contributors from the Center for Christian Virtue, Independent Women’s Forum, the Ethics and Public Policy Center, and clinical practice. They include Aaron Baer, Victoria Coley, Neeraja Deshpande, Eithan Haim, Jordan Jantz, Aaron Kheriaty, May Mailman, Andrea Mew, Rachel Morrison, and Quentin Van Meter. The report describes itself as noncomprehensive and points readers to a separate HHS evidence review for its full clinical assessment of pediatric gender medicine. ^[1],[5]

Share

Its thesis is broader than insurance fraud. HHS says financial incentives attracted hospitals, gender ideology shaped clinicians and medical organizations, Biden-era policy pressured providers and payers to expand access, and weak oversight left families with few brakes on the process. The report presents the billing data as the most measurable part of a much larger institutional failure. ^[1]

A lifelong revenue stream

HHS traces the expansion from a handful of specialized clinics in the early 2010s to programs at more than 225 hospitals and health systems by the early 2020s. The report starts with the economics of pediatrics: lower physician compensation, thinner departmental margins, and less institutional influence than high-revenue adult specialties. Gender clinics, HHS says, changed that calculation by bringing endocrinology, adolescent medicine, plastic surgery, urology, and gynecology into a continuing chain of reimbursable care. ^[1]

The report calls this the ‘captive patient’ model. Endocrine Society and WPATH guidance called for clinical assessments every three to six months, periodic laboratory testing, bone-density scans, and continuing multidisciplinary follow-up. HHS interprets that schedule as a predictable stream of office visits, prescriptions, facility fees, monitoring, and later procedures extending into adulthood. ^[1]

The cost estimates cited by HHS are concrete. Annual payer spending averages about $545 for androgens, $735 for estrogens, and $16,385 for GnRH agonists. The report cites mean health-plan payments of $12,680 for mastectomy with chest masculinization, $17,426 for mammoplasty, $53,645 for vaginoplasty, and $133,911 for phalloplasty. It places lifetime spending at $25,000 to $75,000 without surgery and says combined operations can push the surgical total toward $170,000. Fertility preservation, voice therapy, hair removal, complication care, and later restorative treatment add more. ^[1]

At the institutional level, the report cites the Stop the Harm database for $119,791,202 in submitted charges since 2019, 5,747 minors recorded as having undergone surgeries, and 8,579 minors recorded as receiving hormones or puberty blockers. The same database placed Mount Sinai at more than $8.2 million and Boston Children’s Hospital at more than $6.5 million in billed charges from 2019 through 2023. HHS presents these figures as evidence that pediatric gender programs became meaningful revenue lines inside large health systems. ^[1]

How HHS says providers secured payment

The report’s second chapter centers on ICD diagnosis codes. CMS requires claims to describe the documented condition at the highest available specificity. HHS argues that some providers substituted physical-disease codes for gender-related diagnoses so that insurers, Medicaid, or Medicare would pay for puberty blockers and hormones. The report focuses on E34.9, endocrine disorder unspecified, and E30.1, central precocious puberty. ^[1],[4]

HHS cites earlier public evidence that E34.9 had become a proxy code. A 2016 CMS-linked study described the predecessor endocrine code as a way to avoid the stigma of a gender-identity diagnosis. A 2023 University of Iowa analysis used E34.9 to help identify gender-expansive patients because the code sometimes appeared in place of a gender-dysphoria code. The report interprets these practices as evidence that code substitution was known within the field. ^[1]

HHS then tested the pattern in nationwide all-payer claims from 2015 through 2025. Its first cohort identified puberty-blocker claims for patients ages 9 through 17 carrying an E34-family diagnosis while excluding claims containing an F64 gender diagnosis or E30.1 precocious-puberty diagnosis. HHS reports nearly $50 million in submitted charges across that cohort, including more than $40 million tied specifically to E34.9. The HHS press release summarizes the figure as approximately $50 million associated with E34.9; the report’s detailed analysis assigns more than $40 million to that specific code. ^[1],[2]

The second cohort identified nearly $11 million in puberty-blocker claims for patients ages 13 through 17 carrying E30.1. HHS notes that central precocious puberty ordinarily begins before age 8 in girls or age 9 in boys and that blocker treatment usually ends around ages 10 or 11. The report therefore treats claims carrying that diagnosis in teenagers as evidence of potentially false coding. ^[1]

HHS states its data limits in the appendix. Patient ages were estimated from birth year using July 1 as a default birthday. Dollar totals are submitted charges, and some claims carry null amounts. Organizations may appear under more than one spelling, and inclusion in the roster means only that an organization billed at least one qualifying claim. The report calls the results ‘directional signals rather than findings’ and sends investigators to the underlying records for verification. ^[1]

HHS says the coding practice was openly promoted

The report moves from claims patterns to statements by people and organizations in the field. It cites a 2021 WPATH training presentation that listed E34, E34.9, and hypogonadism codes in materials on insurance coverage and adolescent care. It cites QueerDoc’s public explanation that some providers use E34.9 for laboratory work, prescriptions, and visits. It also reproduces Planned Parenthood of Southeastern Pennsylvania’s statement that the organization typically used E34.9 and used a gender-identity code when necessary. ^[1]

HHS treats these statements as evidence of a billing workaround operating in plain sight. The report also recounts whistleblower allegations at Texas Children’s Hospital, state lawsuits alleging use of endocrine-disorder, precocious-puberty, and contraception codes, and federal inquiries seeking communications about alternative diagnoses. It connects those allegations to a May 2026 resolution under which Texas Children’s Hospital agreed to pay $10 million and create a clinic for detransitioners, and to a June 2026 Cleveland Clinic resolution that included $2 million for restorative care. ^[1]

The report recommends periodic Medicaid reviews of these codes, comparable controls inside managed-care organizations, and referral of suspicious claims to Medicaid Fraud Control Units, state attorneys general, U.S. attorneys, and other law-enforcement bodies. It also points to False Claims Act litigation by private relators as another route for recovering improper payments. ^[1]

The Biden administration supplied federal pressure

Chapter 3 argues that the Biden administration converted affirmation into federal policy. HHS begins with President Biden’s January 2021 executive order on gender-identity discrimination. It then traces the policy through the HHS Office for Civil Rights, which interpreted Section 1557 of the Affordable Care Act to reach refusals of treatment or insurance coverage. Federal courts later vacated the gender-identity portions of the 2022 guidance and the corresponding provisions of the 2024 rule. ^[1]

The report cites CMS approval of Colorado’s essential-health-benefits benchmark, which required coverage of a broad range of interventions, and the CMS administrator’s invitation for other states to follow. It cites a March 2022 Office of Population Affairs document endorsing blockers, hormones, chest surgery, and genital surgery for young people, even though the Food and Drug Administration had approved neither blockers nor cross-sex hormones for gender-related use. It also recounts federal employee coverage directives, litigation positions, grants, and agency programs that HHS says normalized and financed the model. ^[1]

The report’s point is institutional. Hospitals encountered a federal government that treated access as a protected policy objective through civil-rights enforcement, coverage directives, official clinical messaging, federal spending, and legal intervention. HHS says those policies rewarded expansion and discouraged resistance inside medical systems. ^[1]

Medical societies manufactured consensus

Chapter 4 says professional-society bias carried the same ideology into hospital practice. HHS describes extensive overlap between WPATH and the authors of Endocrine Society guidance. It points to recommendations allowing hormones before age 16 in some cases while acknowledging minimal published research below roughly age 14. The report also faults the guideline process for failing to disclose whether authors or affiliated clinicians stood to benefit from wider use of the interventions. ^[1]

HHS makes a similar argument about the American Psychological Association and the American Academy of Pediatrics. It reports that at least seven members of the APA’s eight-person task force were WPATH members or affiliates. It says the AAP’s 2018 policy developed from work involving the Human Rights Campaign, was approved by a 13-member board without a vote of the organization’s 67,000 members, and was reaffirmed in 2023 before the AAP completed its own systematic evidence review. ^[1]

The report says these organizations created the appearance of settled medical consensus, which hospitals then converted into protocols, referrals, consent materials, and standards of care. HHS presents professional endorsement as a mechanism of institutional capture: a small group writes guidance, large associations adopt it, and hospitals treat the resulting policy as authoritative. ^[1]

The people HHS put on the record

The final substantive chapter presents accounts from Sydney Aviles, Lily Burns, Rose Marie, Clementine Breen, Layla Jane, Christy Davidson, Soren Aldaco, and Luke Healy. HHS uses their experiences to show how vulnerable young people entered treatment, how clinicians handled doubt and co-occurring distress, what physical consequences followed, and how little organized care existed for patients who wanted to reverse course. ^[1]

Sydney Aviles describes beginning testosterone, paying $10,000 for a double mastectomy at 18, developing thyroid dysfunction and an elevated red-blood-cell count, and stopping testosterone when vasovagal syncope made therapeutic blood draws untenable. Soren Aldaco describes testosterone, mastectomy, a severe postoperative hematoma, continued medicalization despite worsening symptoms, and no structured detransition plan when she stopped. Luke Healy describes online exposure beginning at 10, estrogen treatment after turning 18, a surgical consultation quoting about $200,000 for facial feminization, and lasting gynecomastia after detransition. ^[1]

These accounts supply the human argument behind the claims analysis. HHS portrays an organized pathway into treatment and an improvised pathway out. Referrals, approvals, prescriptions, surgery, and billing were available; reassessment, long-term outcome tracking, and restorative care were scarce. ^[1]

What HHS wants done

The report calls for federal and state review of coding and billing, stronger program-integrity controls, referral of suspicious claims, protection for whistleblowers, and continued scrutiny of the clinical practices and policies surrounding pediatric gender medicine. The August 13 referrals put that recommendation into motion. Vance sent the provider list to DOJ. Kennedy sent an objectively defined claims cohort to OIG. ^[1],[2],[3]

For the clinical evidence, Wolves in White Coats points to HHS’s separate Treatment for Pediatric Gender Dysphoria review. The policy report uses that evidence judgment as its premise: hospitals expanded irreversible and long-duration interventions while safety and efficacy remained inadequately established. Its own contribution is the institutional account of how money, coding, federal policy, professional authority, and ideology worked together. ^[1],[5]

HHS closes by saying the potential fraud is only the ‘tip of the iceberg.’ That is the case the department has placed before DOJ, OIG, state Medicaid programs, insurers, hospitals, and the public. The report names the revenue model, the diagnostic codes, the institutions, the policy machinery, and the patients who say they were harmed. The investigation now begins where the report ends: inside the claims and medical records. ^[1],[2],[3]

Leave a comment

Sources

1. HHS, Wolves in White Coats: How Doctors and Hospitals Pushed and Profited from the Fraud of ‘Gender Medicine’

2. HHS release announcing Wolves in White Coats and the federal referrals

3. Secretary Kennedy’s referral letter to the HHS Office of Inspector General

4. CMS, FY 2025 ICD-10-CM Official Guidelines for Coding and Reporting

5. HHS, Treatment for Pediatric Gender Dysphoria: Review of Evidence and Best Practices

Anthony Fauci does not appear in The Dog Stars. He is nevertheless in the theater.

Ridley Scott’s new science-fiction film opens after a catastrophic flu has erased most of humanity. A pilot, his dog, and an armed former Marine inhabit what remains. The machinery is familiar: contagion, social collapse, desperate survivors, violent scavengers, and one more search for a pocket of human decency after civilization has failed. Scott did not invent this machinery, but the director of Alien and Blade Runner is one of the people who taught modern audiences how to surrender to it.

This time, audiences largely stayed home. The Dog Stars opened in fifth place with about $8 million in North America and another $11.3 million overseas, against a reported production budget of $70 million to $100 million. The Associated Press called it a flop. The result was especially poor because Boxoffice Pro had forecast an $18 million to $23 million domestic opening. The film landed at less than half the low end of that range.

There are ordinary explanations. Critics found the movie slow and derivative. As of September 1, Rotten Tomatoes showed a 39 percent critic score and a 71 percent verified-audience score. Marketing was weak. Adult dramas have a difficult theatrical market, streaming has trained people to wait, and post-apocalyptic fiction has been feeding on itself for years. A bad opening weekend cannot tell us what millions of people privately thought.

Yet the scale of the miss invites a harder question. Has the public’s experience of COVID changed the mental contract on which the killer-virus movie depends? More bluntly, did Fauci and the institutions for which he became the human logo spend so much public trust that even Ridley Scott can no longer borrow it for two hours?​

The genre ran on institutional credit

The classic outbreak story asks for a larger gift than belief in an unusually lethal pathogen. It asks the audience to accept a hierarchy of knowledge. The virus is invisible. The scientist can see it. Government may be slow, corrupt, frightened, or politically compromised, but somewhere inside the system there is a technically competent truth-teller trying to get reality through the door. The audience is expected to recognize that person immediately, usually by the white coat, the lab bench, or the lonely briefing at which the room finally falls silent.

The formula survived many variations because it rested on a presumption: the expert might be wrong, but the expert was trying to find out. Institutional failure came from delay, cowardice, budget cutting, or a villain who ignored the scientists. Once the pathogen was loose, emergency orders carried a certain moral clarity. People who resisted quarantine were reckless. People who demanded ordinary liberties while bodies accumulated were selfish. The drama could compress epidemiology and ethics into the same clean line.

COVID broke that line.

The public watched recommendations change, which is normal in a developing emergency. It also watched provisional judgments presented as settled science, policy preferences presented as epidemiological necessities, and dissent treated as evidence of defective character. Those are different things. Science can revise itself without shame. Authority cannot demand obedience on the ground that uncertainty has ended and then claim the privileges of uncertainty when the rule fails.

That distinction became painfully concrete. In his 2024 congressional interview, Fauci said the six-foot distancing rule “sort of just appeared” and that he was not aware of studies supporting that exact threshold. He did not say distancing itself was useless; the admission was narrower and more revealing. A number with enormous social consequences had acquired the tone of scientific law without the evidentiary pedigree the public had been led to assume.

The same House transcript records Fauci saying that he technically signed off on NIAID grants without reviewing the individual proposals and that he could not say with confidence how the biosafety of foreign laboratories receiving NIAID funds was independently verified.

These details matter to cinema because the outbreak genre relies on compression. A script needs the audience to accept that “the scientists say” is enough to move the plot. COVID trained a large part of the audience to ask which scientists, reviewing which data, operating under which definition, with what conflicts, and under whose institutional protection. That is terrible for a lazy screenplay. It is healthy for a republic.

The expert is no longer an innocent narrator

Fauci became the face of the American pandemic response because he was made to carry more than scientific advice. He became a character: reassuring, unflappable, credentialed, above the political noise, and able to translate an uncertain emergency into instructions. Supporters treated attacks on him as attacks on expertise itself. Opponents increasingly saw a powerful official whose public certainty exceeded what the record could support.

The conflict has not faded into history. On August 6, 2026, only three weeks before The Dog Stars opened, a Republican-led Senate committee voted along party lines to hold Fauci in contempt after he invoked the Fifth Amendment more than 100 times during questioning about the pandemic. Senator Rand Paul then sent a referral to the Justice Department without a vote of the full Senate.

Fauci’s attorney called the proceeding a political stunt, Democrats challenged the process, and one lawyer told Reuters that the referral could be legally ineffective without full-Senate approval. Those disputes matter legally. Culturally, the picture is already fixed: the man sold to the public as the calm voice of scientific transparency was now declining to answer questions under a constitutional protection against self-incrimination. Reuters documented both the vote and the dispute over its validity.

Invoking the Fifth destroys the old Hollywood shorthand in which the senior infectious-disease official can be presumed to function as the story’s disinterested narrator. A post-COVID audience may reasonably wonder whether the expert on screen is explaining the emergency, managing liability, protecting a research program, shaping a politically useful narrative, or doing several of those things at once.

That question now arrives before the monster. And the public demands its own skepticism be heard.

Why viewers may withhold the gift of suspended disbelief

Suspended disbelief is often described as something an audience does for itself. The phrase makes the transaction sound passive, as if viewers simply turn off their skepticism at the door. Good storytelling earns that concession. The viewer lends the filmmaker a set of assumptions so the film does not have to prove every brick in its world.

The killer-virus movie once received several assumptions almost free of charge. A natural spillover was an adequate origin story. A public-health agency was presumed to be working from the best available evidence. A changing recommendation meant the evidence had changed. Suppression of a claim suggested that the claim was dangerous nonsense. The pharmaceutical countermeasure entered the story as rescue, while the people delaying its distribution supplied the conflict. Coercive rules were regrettable necessities whose moral legitimacy came bundled with the emergency.

Millions of people no longer grant those assumptions automatically.

First, the origin of a pandemic can no longer be treated as atmospheric background. COVID taught the public about gain-of-function disputes, grant intermediaries, biosafety levels, unpublished databases, intelligence assessments, and the institutional stakes of the lab-leak question. Fauci himself acknowledged in 2024 that a laboratory accident is a possibility and not inherently a conspiracy theory, while saying the evidence he had seen still favored natural emergence.

A film that begins with “a flu wiped out nearly everyone” and rushes directly to armed scavengers may feel as if it has skipped the part of the story the audience now finds most urgent: Who made this possible, who knew what, and who controlled the first account of what happened?

Second, the public has learned to separate a scientific observation from a policy command. A pathogen may spread through the air; it does not follow that every closure, distance rule, school policy, mandate, passport, or employment consequence was dictated by a laboratory result. Each policy contains value judgments about risk, age, proportionality, collateral damage, and who must bear the cost. Pandemic movies often erase those judgments by placing a scientist in front of a graph. After COVID, the graph no longer completes the argument.

Third, risk is no longer a single red number. The public spent years hearing about age gradients, comorbidities, prior infection, waning immunity, variant-specific behavior, absolute and relative risk, and the difference between preventing severe disease and preventing transmission. Viewers do not need graduate training in epidemiology to notice when a script uses “deadly flu” as a magic solvent that dissolves demography, immunity, treatment, logistics, and time. The more biologically sweeping the premise, the more explanation it now needs.

Fourth, the memory of coercion changes the moral casting. Before 2020, the citizen who challenged an emergency order could be written as the fool who opens the gate. After school closures, business failures, employment ultimatums, travel restrictions, and the policing of family life, some viewers will inspect the order before condemning the dissenter.

They will ask whether the intervention has evidence, whether the burden is distributed fairly, whether less destructive options were considered, and whether the official issuing it will pay any price for being wrong. Hollywood’s old moral shortcut no longer works on everyone.

Fifth, the audience has seen information control become part of the emergency response. The lab-leak hypothesis, the durability of naturally acquired immunity, the limits of vaccination against transmission, the merits of particular treatments, and the harms of prolonged school closure were not argued in a quiet seminar.

Claims were labeled, throttled, moralized, or attached to political identities. Some claims deserved rejection. Others deserved open examination. The lasting injury came from the attempt to make institutional permission a substitute for scientific adjudication. A movie that assigns truth to the credentialed character and error to the outsider now asks viewers to forget the structure they just lived through.

Sixth, a pandemic is no longer safely imaginary. For many people it means a parent who died alone, a child who lost part of a childhood, a business that never reopened, a delayed cancer diagnosis, a hospital rule that prevented a final visit, or a career damaged by a mandate. The apocalypse may still entertain, but it no longer arrives as a clean thrill. A fictional flu that exists only to clear the landscape for a handsome survivor can feel less like daring speculation than appropriation of unresolved experience.

Finally, the real pandemic made bureaucracy more frightening than the virus story Hollywood prefers to tell. The old plot locates danger in the microbe and heroism in the response. The post-COVID plot must allow a more disturbing possibility: institutions can be partly competent, partly self-protective, scientifically informed in some decisions, politically captured in others, and unable to admit which is which. That is a richer story. It is also much harder to make.

Fauci is a face, not a box-office model

None of this means Anthony Fauci personally removed $10 million from Ridley Scott’s opening weekend. The box office does not provide a referendum question with every ticket. The Dog Stars may have failed because people heard it was dull. Its 39 percent critic score and middling audience response provide plenty of explanation. The same weekend, viewers spent heavily on Spider-Man, Christopher Nolan’s The Odyssey, and a rescued Wile E. Coyote film. The audience has not abandoned theaters, fantasy, danger, or cinematic unreality.

Nor has every outbreak-adjacent story failed since COVID. Audiences will still follow zombies, fungi, aliens, radiation, engineered organisms, and social collapse when the characters and world earn their attention. The important change is that a pandemic can no longer do all the narrative work by itself. “A virus wiped out civilization” used to be a premise. It may now sound like a writer asking to skip the audit.

Fauci matters because public memory condenses systems into people. He did not act alone, set every policy, write every headline, run every platform, or make every hospital decision. He became the emblem of an alliance among federal health agencies, political executives, media organizations, universities, technology companies, and pharmaceutical interests that often spoke with one voice while insisting that the voice belonged to science.

The collapse of trust in that arrangement is measurable. Pew Research Center reported in January 2026 that 77 percent of Americans retained at least a fair amount of confidence in scientists, which is still high, but below the 87 percent recorded in April 2020. The partisan division is now severe: 90 percent of Democrats expressed at least fair confidence, compared with 65 percent of Republicans.

That is not the disappearance of trust. It is the disappearance of a common presumption. A national movie audience can no longer be expected to bring the same view of scientific authority into the theater.

Hollywood has to earn the premise again

The killer-virus film is not dead. Its easy version may be.

The next great pandemic movie will have to understand that knowledge has a production history. Someone funds the work. Someone selects the model. Someone decides which signal counts as an emergency. Someone chooses the comparison group, writes the press release, defines the acceptable hypothesis, and determines what the public is permitted to hear before the evidence is complete. Those choices do not make science fraudulent. Hiding them makes scientific authority brittle.

A serious outbreak film made after COVID should be willing to show uncertainty without making uncertainty a synonym for incompetence. It should distinguish evidence from policy, and policy from enforcement. It should let scientists disagree for substantive reasons. It should give the audience enough biology to understand why risks differ, enough institutional detail to see why incentives matter, and enough respect to recognize that citizens who ask for evidence are not automatically villains.

That approach would also restore suspense. The most frightening question is no longer whether the official will persuade the president to act before the virus escapes. It is whether the people controlling the response can tell the truth when the truth threatens the institutions, programs, and reputations on which their authority rests.

Fauci did not destroy Ridley Scott. He may have helped destroy the automatic credibility of the world in which the classic killer-virus movie took place. That world required an audience willing to assume that the pathogen was opaque, the experts transparent, and emergency power temporary. After COVID, each assumption looks like a claim to be tested.

Hollywood can still ask for suspended disbelief. The audience is simply checking the borrower’s books.

Subscribe now

Share

Leave a comment

The August 23 article, “mRNA Cancer Vaccines: Promising but Dogged by Misinformation”, celebrates Moderna and Merck’s personalized melanoma treatment while dismissing concern about cancer following COVID-19 vaccination. CNN’s author presents the public with a false binary: either embrace mRNA cancer therapeutics or join the irrational people allegedly threatening scientific progress.

That is advocacy disguised as medical reporting.

Personalized therapeutic cancer vaccines may prove valuable. That possibility does not validate every mRNA product, every lipid nanoparticle formulation, every encoded antigen, every dosing schedule, or every use of the platform. A therapeutic product designed from an individual patient’s tumor and administered to a patient with resected high-risk melanoma is not scientifically interchangeable with a standardized prophylactic product administered repeatedly to hundreds of millions of healthy people.

The word “mRNA” does not confer class-wide safety or class-wide efficacy. Payload, sequence, nucleoside chemistry, lipid composition, dose, route, biodistribution, antigen persistence, manufacturing quality, patient selection, concomitant therapy, and clinical context all matter. The benefit-risk threshold for a patient facing recurrent melanoma also differs radically from the threshold for a healthy child, adolescent, or young adult.

CNN’s article erases these distinctions because acknowledging them would dismantle its central narrative.

CNN declared success before the public had seen the Phase 3 results

CNN describes the intismeran program as a success. Yet Moderna and Merck had not publicly released the detailed Phase 3 results when CNN published its article. The companies announced that the trial met its recurrence-free-survival and distant-metastasis-free-survival endpoints, but they did not disclose the effect estimates, absolute event rates, Kaplan-Meier curves, subgroup results, mature overall-survival data, or a complete adverse-event analysis. The detailed findings were reserved for a future medical meeting.

That is not a minor omission. Those data determine whether the result represents a clinically important advance, a modest delay in recurrence, a benefit confined to a subgroup, or an effect purchased at an unacceptable toxicity burden. The treatment was also tested in combination with pembrolizumab. Attribution therefore requires examination of interaction, treatment discontinuation, immune-mediated toxicity, and exposure duration. A corporate announcement that endpoints were met cannot answer those questions.

Reuters correctly reported that detailed results had not been provided. CNN nevertheless converted an undisclosed interim dataset into a vehicle for attacking people who request long-term safety evidence. Reuters, August 19, 2026

CNN is demanding public trust in results the public cannot yet inspect.

CNN’s claim about population evidence is demonstrably false

A syndicated reproduction of the CNN report states that “large population studies have found no increased risk of cancer following COVID-19 vaccination.” That categorical statement is false. Two large population-based studies have reported statistically significant positive associations between COVID-19 vaccination and cancer outcomes.

The first examined 296,015 residents of Pescara Province, Italy. Receipt of at least one COVID-19 vaccine dose was associated with a 23% higher adjusted hazard of hospitalization carrying a cancer diagnosis: HR 1.23, 95% CI 1.11–1.37. The reported associations included colorectal cancer, HR 1.35; breast cancer, HR 1.54; and bladder cancer, HR 1.62. Among participants without a recorded prior SARS-CoV-2 infection, the all-cancer estimate increased to HR 1.31. The verified publication is Acuti Martellucci et al., 2025, PMID 40881928, DOI 10.17179/excli2025-8400.

The second used South Korea’s National Health Insurance database, beginning with 8,407,849 individuals. Its authors reported increased one-year hazards for thyroid, gastric, colorectal, lung, breast, and prostate cancers. The journal subsequently posted an editorial notice stating that concerns had been raised and were under investigation. That notice requires caution; it does not permit CNN to pretend that the study does not exist. The correct journalistic statement would be that population studies have produced conflicting and methodologically contested findings that do not yet establish causation. Kim et al., 2025, PMID 41013858, DOI 10.1186/s40364-025-00831-w.

CNN did not critically evaluate these studies. It erased them.

That is not fact-checking. It is evidence selection.

The Italian study was designed to lose cancer events

The Italian study deserves severe criticism—but not for the reason CNN would prefer.

Its analysis was structured in ways that could suppress or obscure a cancer signal. Vaccinated participants entered the cancer risk set only after surviving 90, 180, or 365 days following vaccination without a qualifying cancer hospitalization. Cancer events occurring before the selected landmark were removed from the incident-cancer analysis or reclassified as prior cancer.

That creates a post-exposure cancer-free survival requirement. It excludes the very outcomes that would matter if vaccination accelerated an occult malignancy, triggered recurrence, unmasked a hematologic cancer, or promoted rapid progression during the first months after exposure.

The 365-day analysis intensifies the distortion. Participants must survive and remain free of a qualifying cancer hospitalization for an entire year after vaccination before entering the analysis. The authors then described the result as a reversal of the cancer association. It was no such thing.

At 365 days, the ≥1-dose all-cancer estimate remained above the null at HR 1.13, 95% CI 0.99–1.30. Breast-cancer hospitalization remained significantly elevated at HR 1.63, and bladder-cancer hospitalization remained significantly elevated at HR 1.82. Only the highly selected ≥3-dose all-cancer group produced an estimate below 1. People entering that group had to survive to a third dose, remain eligible and willing to receive it, avoid a qualifying cancer event, and then survive the additional imposed lag.

That is survivor selection and depletion of susceptible individuals—not a clean test of latency.

The study also substituted first cancer hospitalization for cancer incidence. It therefore missed pathology-only diagnoses, outpatient diagnoses, early-stage cancers managed without admission, and cancers treated outside the captured hospitalization system. By excluding people with prior cancer, it also removed recurrence, transformation, and accelerated progression—the clinical phenomena at the center of the disputed safety signal.

Its dose categories overlapped. The ≥1-dose group included the ≥3-dose group, making the comparison incapable of establishing a conventional dose-response relationship. The primary analysis pooled different vaccine platforms, products, schedules, and mixed-dose combinations. Infection classification depended on recorded testing and outcome-related timing, even though testing requirements differed by vaccination status and changed repeatedly. The published Methods, main tables, figure, and supplementary material also contain irreconcilable 90-day and 180-day cohort-entry dates.

Despite this architecture, the study still found significant increases.

CNN and its author were apparently uninterested in asking how large the signal might have been had the investigators retained early events, included recurrence and progression, used complete cancer-registry and pathology data, defined vaccination as a time-varying exposure, aligned participants on calendar time, and eliminated the post-vaccination cancer-free survival gate.

CNN’s disinterest runs in only one direction. Weaknesses in a study reporting harm become grounds for ignoring the signal. Weaknesses in reassuring studies disappear into the phrase “large population studies.”

“Misinformation” has become a method for evading the unresolved question

The phrase “turbo cancer” lacks a standardized clinical definition. That does not settle whether vaccination could accelerate progression, alter tumor immune surveillance, reactivate oncogenic viruses, affect preexisting neoplasia, or interact with particular cancers in susceptible patients.

Rejecting an imprecise popular term does not reject every biological or epidemiological hypothesis placed beneath it.

Likewise, showing that mRNA does not ordinarily integrate into nuclear DNA does not establish that an mRNA-lipid nanoparticle product cannot influence cancer biology through other pathways. Integration is one proposed mechanism, not the entire hypothesis space. Immune dysregulation, altered interferon signaling, inflammatory signaling, antigen persistence, lymphocyte effects, latent-virus reactivation, biodistribution, off-target protein expression, and changes in tumor immune surveillance require product-specific investigation.

CNN replaces this multidimensional inquiry with a single elementary rebuttal: mRNA does not alter the genetic code. That answer addresses the easiest claim to dismiss while leaving the consequential questions untouched.

This is a straw-man defense of a platform, not an investigation of product safety.

The FDA must intervene—but it must regulate evidence, not speech

The FDA should not censor CNN, physicians, patients, or scientists. It should crack down on the biased evidentiary practices that allow manufacturers, researchers, and media organizations to claim that a safety question has been resolved when it has not.

FDA’s Center for Biologics Evaluation and Research and Oncology Center of Excellence should require a formal, independent cancer-safety assessment across authorized and licensed COVID-19 vaccine products. That assessment must link vaccination records to state and national cancer registries, pathology data, outpatient oncology records, hospitalization files, mortality records, and longitudinal exposure histories.

The protocol must be publicly registered before analysis. Vaccination must be modeled as a time-varying exposure. Every participant’s unvaccinated person-time must be retained until the actual date of vaccination. Early post-vaccination events must remain visible in prespecified risk windows rather than being deleted through landmark qualification. Analyses must distinguish new primary cancers, recurrences, progression, transformations, hematologic malignancies, and cancer-specific mortality.

The investigators must stratify by product, lot, dose number, sequence, age, sex, prior cancer, immune status, prior infection, and clinically meaningful latency. They must adjust for cancer screening, healthcare utilization, smoking, socioeconomic status, frailty, and differential testing. They must address death as a competing event and publish negative-control outcomes capable of detecting residual bias.

FDA should require public release of the analytic code, complete variable definitions, cohort flow diagrams, event counts within every exposure window, and sufficient deidentified data to permit independent replication. Contradictory index dates and silent reclassification of post-exposure cancers as “previous cancers” should disqualify an analysis from supporting a regulatory safety claim.

FDA already recognizes that reliability, relevance, prespecification, traceability, and appropriate causal design determine whether real-world data can support a safety conclusion. Its own real-world-evidence framework cannot remain an aspirational document while agencies and manufacturers cite studies containing immortal-time bias, healthy-vaccinee bias, post-exposure conditioning, informative censoring, outcome misclassification, and selective latency windows. FDA Real-World Evidence Program

FDA must also stop allowing evidence from one mRNA product to function as reputational cover for another. A personalized therapeutic cancer vaccine cannot validate a prophylactic COVID-19 vaccine any more than the success of one monoclonal antibody validates every monoclonal antibody. Each product requires its own pharmacology, biodistribution, manufacturing, dose, safety, and benefit-risk assessment.

If a manufacturer or sponsor makes categorical promotional claims that the platform does not increase cancer risk, FDA should demand the evidence supporting that claim and enforce the applicable promotional and labeling requirements. Where commercial advertising falls outside FDA’s jurisdiction, the agency should refer the matter to the Federal Trade Commission. FDA should also convene a public advisory committee containing cancer epidemiologists, causal-inference experts, tumor immunologists, pathologists, pharmacovigilance specialists, affected patients, and qualified scientific critics—not another panel assembled to ratify a predetermined conclusion.

CNN has confused public trust with public submission

Trust does not arise from commanding patients to stop asking questions. It arises when institutions preserve events instead of deleting them, test competing hypotheses, disclose unfavorable analyses, publish code, release data, correct errors prominently, and stop labeling unresolved evidence as misinformation.

CNN’s author did none of that. The article treats confidence in mRNA technology as the desired endpoint and scientific evidence as material to be arranged around it. It asks whether public concern might damage a promising commercial technology. It never seriously asks whether institutional overstatement, biased safety studies, suppressed uncertainty, and incurious journalism created that concern.

The public does not owe an experimental platform its trust.

Researchers owe the public valid study designs. Manufacturers owe regulators complete evidence. FDA owes the country active and independent surveillance. Journalists owe readers an honest account of conflicting findings.

A promising melanoma therapy deserves rigorous evaluation. So do the cancer signals following COVID-19 vaccination. Anyone who supports the first while demanding suppression of the second is not defending science.

They are defending a franchise.

Share

Leave a comment

Pennsylvania officials report two measles-associated deaths in Lancaster County; coroner and state health department dispute cause-of-death classification

Assembled by the IPAK.NEWS Debiasing Engine · no human review · from 25 scored articles · engine ipn-0.1

Pennsylvania Governor Josh Shapiro and state Health Secretary Debra Bogen announced two measles-associated deaths of unvaccinated Lancaster County residents, described as the first such deaths in the state in 35 years. Lancaster County Coroner Stephen Diamantoni subsequently told local media and a county commissioner that his office had no record of measles deaths; one case involved an infant who died from a lacerated spleen and tested positive for measles, but the forensic pathologist did not attribute the death to measles. The state health department maintained that both deaths were measles-associated and that its health secretary had reviewed the case information. HHS Secretary Robert F. Kennedy Jr. said the announcement may have been premature or that the deaths may have been fabricated by a staffer, while Democratic officials and others called for his resignation, citing his history of anti-vaccine rhetoric. The CDC has recorded 2,777 measles cases nationally in 2026, and Pennsylvania has confirmed 393 cases across 28 counties.

Pennsylvania officials announced two “measles-associated” deaths in Lancaster County. They did not release case histories, final causal determinations, death certificates, or enough clinical information for the public to know whether measles was the immediate cause of death, an underlying cause, a contributing condition, or simply a laboratory-confirmed infection present at the time of death.

That uncertainty should have governed every headline and every official statement that followed.

Instead, the political conclusion arrived first.

Governor Josh Shapiro called the deaths “preventable”. The Pennsylvania Department of Health used them to reinforce its vaccination campaign. The American Academy of Pediatrics treated them as deaths from measles. And MedPage Today^[1], a publication written for clinicians, discarded the state’s own qualifying language in the headline and announced: “U.S. Measles Deaths Reported.”

This is precisely the kind of causal inflation that destroyed public trust during COVID-19. A positive test became a diagnosis. A diagnosis became a cause of death. A provisional count became an emotionally charged certainty. Then that certainty was used to demand compliance and marginalize anyone who asked what the underlying records actually showed.

We do not need that type of response again.

Measles can be serious. The MMR vaccine reduces the risk of infection and severe outcomes. Neither fact gives journalists, professional societies, or elected officials permission to convert an unresolved “associated with” finding into a completed “caused by” determination.

“Associated” is not a decorative word

The Pennsylvania Department of Health’s announcement^[2] described two “measles-associated deaths.” That is a surveillance category, not a full causal explanation.

According to local reporting^[3], the department acknowledged that the term means there was laboratory or epidemiological evidence of measles, but not necessarily that measles was the immediate cause of death. One infant’s death remained under investigation by the Lancaster County coroner, with a ruptured spleen reported as the immediate cause. The other death was not presented to the coroner, and officials disclosed no case-specific clinical explanation.

Those facts do not prove that measles played no causal role. They prove that the role had not been publicly established.

That distinction is elementary in medicine. A patient may die with an infection, from an infection, or from a chain of events in which the infection is one contributing factor. A serious news report would explain those categories and state what evidence remained unavailable. A responsible government would release a de-identified clinical summary once the investigation was complete.

Pennsylvania instead supplied a label broad enough to support surveillance and vague enough to support a political campaign.

MedPageToday erased the uncertainty

MedPage Today had one central job: tell its clinical readership what had and had not been established. It failed.

Its headline said “U.S. Measles Deaths Reported.” The first sentence quietly restored the missing qualifier and referred to “measles-associated deaths.” The article never explained why that qualifier mattered. Certainty occupied the headline; uncertainty was buried in the body.

That is not a minor editing choice. It changes the claim.

The article then called 2026 “record-breaking,” only to report later that the national count was the highest since 1991, when nearly 10,000 cases were recorded. A year can be the highest in thirty-five years without being an all-time record. “Record-breaking” without the relevant limit is alarm language, not precision.

The same slippage appeared in the vaccination statistics. MedPage characterized the year’s cases as overwhelmingly “unvaccinated,” while the cited 94 percent category combined people known to be unvaccinated with people whose vaccination status was unknown. Unknown is not a synonym for unvaccinated. Combining the categories may be legitimate in a surveillance table; rhetorically converting the combined figure into a known-unvaccinated figure is not.

Then came the political package. The report pivoted from the two unresolved Pennsylvania deaths to Robert F. Kennedy Jr., President Trump, autism claims, and a statement about the MMR vaccine. Whatever one thinks of those figures or their statements, none of that material determines the cause of either Lancaster County death. Its placement supplies villains before the medical record supplies causation.

This is advocacy by arrangement. Put the frightening headline first, add grieving authorities, collapse “unknown” into “unvaccinated,” insert national political antagonists, and close with the possible loss of measles-elimination status. The reader is pushed toward a verdict that the case-specific evidence has not yet earned.

MedPage Today is not a general-interest tabloid. It markets itself to physicians and health professionals. Its audience understands—or should understand—the difference between a laboratory finding, a surveillance classification, an underlying cause, and an immediate cause. Clinicians deserved analysis. They received press-release amplification.

Andrew Racine should know better

The same standard applies to Andrew Racine, MD, PhD, president of the American Academy of Pediatrics. MedPage quoted him calling it “heartbreaking to witness deaths from a disease that is preventable” by vaccination.

The heartbreak is not in dispute. The causal declaration is.

Racine did not cite a death certificate, autopsy, coroner’s finding, or case review establishing that measles caused either death. Yet his statement moved directly from two “measles-associated” reports to deaths from a vaccine-preventable disease. As a physician speaking for a major professional society, he should be more disciplined than a campaign surrogate.

Words such as “heartbreaking” have a legitimate human place after tragedy. They cannot substitute for a causal analysis. Emotion should follow the evidence, not be deployed to preempt questions about it.

Pediatricians should express the same concern following vaccine-caused deaths and injuries. They tend not to.

The AAP could have said that measles can cause severe illness, that vaccination is strongly protective, and that the organization awaited completed reviews of the Pennsylvania cases. That would have been accurate, humane, and useful. Instead, Racine supplied the emotional certainty that the state’s unresolved classification did not.

This is how institutional credibility is spent: not in one spectacular lie, but in a series of small overstatements defended as necessary for the public’s own good.

COVID-19 already taught this lesson

During COVID-19, officials repeatedly blurred distinctions that mattered. A death following a positive test could enter public discussion as a COVID death without the public seeing the clinical chain. Provisional numbers were treated as final when they supported urgent action, then revised quietly. Questioning classification rules was often characterized as denial rather than quality control.

The lesson is not that COVID-19 was harmless or that infectious-disease surveillance is illegitimate. The lesson is that emergency communication becomes propaganda when authorities conceal uncertainty while broadcasting conclusions.

Vaccine programs rely on on voluntary cooperation. Cooperation depends on trust. Trust depends on officials stating what they know, what they infer, and what they do not yet know. Once those categories are collapsed, every later correction looks like a retreat and every demand for compliance sounds like manipulation.

Pennsylvania and MedPage Today are repeating the category error now. “Measles-positive,” “measles-associated,” and “caused by measles” are being treated as interchangeable because the stronger wording produces a stronger emotional response.

That response is politically useful. It is not scientifically defensible without the underlying evidence.

The regulation waiting behind the headlines

The timing matters because the Shapiro administration was simultaneously advancing Regulation 10-242, IRRC No. 3490^[4], a 502-page proposed rewrite of 28 Pa. Code Chapter 27. The proposal was published on August 8, and public comment remained open through September 21.

Seventeen days after publication, two unresolved measles-associated deaths became national news and a vehicle for categorical political messaging.

Timing alone does not prove that the deaths were orchestrated or that the regulation was invented in response to them. The Department of Health has contemplated a comprehensive Chapter 27 revision for years; a state regulatory agenda^[5] documented such plans as early as 2014. The proposed package also reached the Independent Regulatory Review Commission before the August death announcement.

Those facts should be stated plainly because a credible critique does not need an invented conspiracy.

The legitimate concern is more direct: officials used unresolved cases to intensify the atmosphere in which an extraordinary expansion and reorganization of communicable-disease authority was being considered. The proposal expressly responds to legal limits exposed during COVID-19, and it does so by broadening statutory language that state courts found insufficient for a statewide school-mask mandate.

The timing warrants scrutiny. The regulation warrants opposition or substantial revision. Neither conclusion depends on proving secret coordination.

A rule written around the COVID defeat

The proposed regulation^[6] does more than update terminology.

It expands the list of reportable diseases, infections, and conditions from roughly 52 to 125, including COVID-19 and an open-ended category for an emerging disease or condition. It creates broader immunization-reporting obligations, subject to a written opt-out. It specifies access to patient medical records and electronic copies. It provides for entry into apartments, buildings, health-care facilities, schools, colleges, universities, and other locations during investigations upon presentation of agency credentials. It requires schools to provide officials reasonable and timely access to students and permits private interviews in circumstances defined by the rule.

Some of these authorities have predecessors in the existing Chapter 27. Existing Chapter 27^[7] already gives health authorities significant investigatory powers, including access to records and authority to seek entry. The fair objection is not that every power appears from nowhere. It is that the proposal consolidates, enlarges, and normalizes these powers while adding broad new language that can be activated far beyond a single measles outbreak.

The most revealing revision appears in § 27.60. The department proposes authority to order measures for the “prevention, containment or mitigation” of communicable disease. In its own preamble, it links that amendment to Corman v. Acting Secretary of the Pennsylvania Department of Health, the case that invalidated the statewide school-mask order because the department had exceeded the authority supplied by existing law.

In other words, the COVID-era legal defeat is not an incidental historical reference. It is part of the rationale for rewriting the rule.

The preamble identifies examples such as restricted access, cleaning, disinfection, and social distancing. The operative language is broader than those examples. It would help rebuild, through administrative rulemaking, authority that proved legally inadequate during the pandemic.

That is a major policy choice. It should not be smuggled through as a technical modernization, and it should not be debated under the emotional pressure of two deaths whose causal status remains unresolved.

The measles provisions deserve their own examination

The proposal also revises Pennsylvania’s measles-control provisions. It extends an exclusion period from 14 days to 21 days for people the state treats as susceptible in an outbreak setting, with return tied to evidence of immunity, vaccination, or the passage of time without additional cases.

During a genuine outbreak, temporary exclusion can be a proportionate infection-control measure. But the rule shifts practical burdens onto families required to prove they are not susceptible or accept vaccination to regain access. In communities with religious or philosophical objections, that power is not a minor administrative adjustment. It can function as coercion even when officials describe vaccination as technically voluntary.

The draft also appears to reverse the existing measles birth-date cutoff, changing “born after December 31, 1956” to “born before December 31, 1956.” If that language is a drafting error, the department should correct it publicly. A 502-page rule that expands coercive authority cannot be waved through on the assumption that consequential errors will be fixed later.

Privacy does not require causal opacity

Pennsylvania officials cite patient privacy when declining to release details. Privacy is a real obligation. It does not require silence about the basis for a public causal claim.

The department can release de-identified information sufficient to support its messaging:

• age range rather than exact age;

• timing of symptoms, testing, hospitalization, and death;

• test type and specimen date;

• immediate, underlying, and contributing causes once determined;

• whether the determination came from a treating physician, medical examiner, or coroner;

• relevant clinical complications without identifying the patient; and

• whether vaccination status was documented or inferred.

If officials want the public to change behavior because of two deaths, they owe the public more than an emotionally potent label.

The same is true of MedPage Today. It should have asked for those facts before promoting an unqualified causal headline. If the state declined, that refusal should have been the center of the story.

What a proportionate response looks like

Rejecting hype is not the same as rejecting public health. A measured response to measles would include prompt identification and isolation of confirmed infectious cases, focused contact tracing, protection of infants and immunocompromised people, access to vaccination and appropriate post-exposure measures, transparent outbreak data, and completed case reviews before categorical death claims are made.

None of that requires a national morality play. None requires treating every unknown vaccination status as unvaccinated. None requires dragging unrelated presidential rhetoric into a case report. And none requires granting the executive branch a 502-page blank check.

The public does not need another campaign in which uncertain deaths become symbols, symbols become panic, and panic becomes leverage for permanent authority.

Measured public health is not inaction. It is action matched to evidence.

Withdraw the rule or rewrite it substantially

Regulation 10-242 should be withdrawn or substantially revised before it advances.

At minimum, Pennsylvania should:

• remove or narrowly define the open-ended “emerging disease or condition” category;

• require judicial authorization for nonconsensual entry except in a documented, immediate emergency;

• require parental notice and clear legal authority before private interviews with minors;

• limit medical-record access to records directly relevant to a defined investigation;

• define objective thresholds, duration limits, review procedures, and appeal rights for exclusion and quarantine;

• separate disease-specific technical updates from broad grants of emergency authority;

• correct the apparent measles birth-date error;

• publish a redline against the current regulation; and

• explain precisely which provisions are intended to overcome the limits recognized in Corman.

The General Assembly should also ask whether authority of this breadth belongs in an agency rule at all. Powers affecting entry, privacy, education, bodily autonomy, and movement deserve explicit legislative debate.

The demand is simple: earn the conclusion

If completed medical reviews show that measles caused or materially contributed to these deaths, Pennsylvania should publish de-identified evidence and say so. If the role was incidental or uncertain, officials and news organizations should correct the record with the same prominence they gave the original claim.

MedPage Today should revise its headline now. Andrew Racine and the AAP should stop speaking as though the case reviews are complete. Governor Shapiro and his health department should separate advocacy for vaccination from the unresolved facts of individual deaths.

COVID-19 taught the public what happens when authorities exaggerate certainty, enlist the press, and treat questions as obstacles to compliance. The result was not durable trust. It was institutional damage that will outlast the pandemic.

Pennsylvania should not repeat that failure. A measles-positive test is evidence. A measles-associated death is a surveillance finding. A death caused by measles is a medical conclusion.

Public officials and medical journalists do not get to erase the distance between them simply because fear closes the gap faster.

Sources and primary documents

[1] MedPage Today. U.S. Measles Deaths Reported. Accessed August 27, 2026.

[2] Pennsylvania Department of Health. Pennsylvania Department of Health Confirms Two Measles-Associated Deaths. Accessed August 27, 2026.

[3] CBS 21 / Local 21 News. Lancaster County reporting on the two measles-associated deaths and the unresolved causal findings. Accessed August 27, 2026.

[4] Pennsylvania Independent Regulatory Review Commission. Regulation 10-242 / IRRC No. 3490 docket. Accessed August 27, 2026.

[5] Pennsylvania Bulletin. 2014 Regulatory Agenda documenting plans for a comprehensive Chapter 27 revision. Accessed August 27, 2026.

[6] Pennsylvania Department of Health. Proposed Regulation 10-242: Communicable and Noncommunicable Diseases. Accessed August 27, 2026.

[7] Pennsylvania Code. Current 28 Pa. Code Chapter 27: Communicable and Noncommunicable Diseases. Accessed August 27, 2026.

Subscribe now

Leave a comment

Shapiro’s overclaim could hardly have been more categorical: “There is no data, none, that links vaccines to autism.” He used that assertion to dismiss President Trump’s suggestion that reducing or spacing childhood vaccinations would reduce autism. Trump’s prediction was not established by the evidence Shapiro reviewed. If the president presented a schedule change as a demonstrated way to reduce autism, he ran ahead of the science. Shapiro then ran just as far in the opposite direction and called the distance certainty.

That is the problem with Shapiro’s handling of the evidence. He did not merely say that the strongest studies weigh against a large average association between MMR vaccination and autism. That would have been defensible. He said there were no data linking vaccines to autism, treated MMR as interchangeable with every vaccine and the entire schedule, substituted a prevalence-by-severity study that contained no vaccine-exposure variable, repeated the prosecution’s account of Andrew Wakefield without examining the defense, and ignored both the William Thompson record and the basic problem of susceptible subgroups. He then turned his own incomplete review into a lecture about scientific integrity.

The claims nobody is making

No one has proposed that their child developed autism after eating cheese. But hundred and thousands and tens of thousands of parents have observed their child change their developmental trajectory after vaccination. So Ben’s correlations do to equal causation argument misses the initial source of the hypothesis: real-world, empirical observation. I dare Ben to spend two days with moms and hear their stories. I dare him to watch Vaxxed and Vaxxed II, and spend a third day explaining his understanding. Perhaps reading this article alone will help him recognize is logical shortcomings.

No serious person is claiming that the average vaccinated child develops autism. I am certainly not making that claim. Vaccines prevent infections, some of which can kill, disable, or injure developing brains. The existence of those benefits does not establish that every vaccine is incapable of contributing to neurodevelopmental injury in every child under every set of biological conditions. Population benefit, individual adverse events, and causal contribution to a heterogeneous diagnosis are separate questions.

Autism is not a single molecular disease with one cause. It is a diagnostic category containing different developmental histories, genetic liabilities, metabolic states, immune phenotypes, levels of function, comorbidities, and likely causal paths. This is not a “complexity” obfuscation point; I’ve spend years studying the literature and have the receipts on plausible mechanisms. The scientifically serious question is whether a particular exposure, formulation, ingredient, combination, or timing window can contribute to a particular autism phenotype in a biologically susceptible minority. Shapiro answered a different question: whether vaccination produces a large average increase in autism across broad populations. When studies did not find that large average increase, he treated the result as proof that no susceptible subgroup could exist.

That inference does not follow. If an exposure raises risk in a small group but has no effect in most people, the population average can dilute the signal. A very large sample helps with statistical power, but it does not repair exposure misclassification, poorly chosen time windows, outcome heterogeneity, post hoc subgroup decisions, or the failure to measure the biology that defines susceptibility. Family history is useful, but “has an older autistic sibling” is not a complete biological definition of mitochondrial impairment, impaired toxicant handling, ER-stress susceptibility, immune dysregulation, or any other proposed pathway.

This does not mean subgroup hypotheses are automatically true. It means they must be specified, tested, replicated, and subjected to interaction analysis. A subgroup claim invented after looking at the data deserves skepticism. A subgroup signal that survives a prespecified analysis deserves investigation. The cure for a difficult causal question is better design, not the declaration that heterogeneity has been averaged out of existence.

What the strongest negative studies actually show

There is substantial evidence against a large population-level MMR effect, and a fair appraisal should say so plainly, but with strong caveats. The 2019 Danish cohort led by Anders Hviid followed 657,461 children and reported an adjusted hazard ratio of 0.93, with a 95 percent confidence interval of 0.85 to 1.02. The investigators also examined several broad risk strata and found no increased MMR-associated autism risk. The study strongly weighs against MMR as a major population-wide cause. It’s flawed in many ways (as I reported), but fine: It tests the wrong hypothesis.

A 2015 study of 95,727 children found no increased autism risk after MMR vaccination among children with or without an older sibling diagnosed with autism. A 2014 meta-analysis combining cohort and case-control evidence likewise found no association between vaccination and autism or autism spectrum disorder. In 2025, the World Health Organization’s vaccine-safety committee reviewed 31 primary studies published since 2010. Twenty, including the studies judged most methodologically rigorous, found no association; eleven reported possible associations, but WHO judged them to have serious methodological problems and a high risk of bias.

In 2012 IOM had looked at 22 and rejected 17 as flawed and reported that 4/5 they left in their consideration were underpowered. They accepted the negative result anyway. Similarly, WHO therefore was flawed in the reaffirmation of its conclusion that the evidence does not support a causal link.

Institute of Medicine’s 2012 report, Adverse Effects of Vaccines: Evidence and Causality, specifically Chapter 4, “Measles, Mumps, and Rubella Vaccine,” and Conclusion 4.8.

The report’s exact accounting is:

• It reviewed 22 epidemiological studies concerning MMR and autism.

• It excluded 12 because they used passive-surveillance data without an unvaccinated comparison group or ecological comparisons without individual-level data.

• It excluded another 5 controlled studies for what it called “very serious methodological limitations.”

• That left 5 publications contributing to the evidence.

• One of those—Farrington et al. (2001)—was a reanalysis of the same 357-case population used by Taylor et al. (1999). Consequently, the IOM described its conclusion as resting on four independent studies/data sets.

• It then concluded: “The evidence favors rejection of a causal relationship between MMR vaccine and autism.” Official IOM chapter

The 2012 IOM review located 22 MMR–autism epidemiological studies, excluded 17 from its evidentiary weighting, and based its high-confidence conclusion on five publications representing only four independent data sets—one population-scale cohort, one substantial case-control study, one much smaller study with serious limitations, and one 357-case data set reported twice.

Is this “strong evidence”?

These findings matter. While they make a large average causal effect of MMR implausible and sharply constrain any defensible hypothesis, detailed understanding of the studies show they could not have detected an association even it one was present. They also do not establish that every vaccine has been tested against autism, that the childhood schedule as a whole has been tested in every relevant way, that every timing question has been answered, or that every biologically defined subgroup has been examined with enough power and measurement accuracy to exclude a clinically meaningful risk.

The Institute of Medicine made these distinctions more carefully than Shapiro did. Its 2004 review concluded that the epidemiological evidence favored rejection of a causal relationship between MMR and autism. The same report explained why its framework had no category called “establishes no causal relationship”: proving universal absence is not equivalent to showing that the balance of evidence weighs against a proposed relationship. In 2012, when the Institute reviewed DTaP-containing vaccines and autism, it found that the epidemiological evidence was insufficient or absent and concluded that the evidence was inadequate to accept or reject causation. That is neither proof of harm nor proof of safety. It is an evidence gap.

I apologize for the repetition, but I think it’s important: Over that time period, they had to reject 17/22 studies, and lean heavily on negative results from underpowered (too small to detect) studies.

Ben: Is this strongest evidence?

The Institute’s 2013 review of the childhood schedule was equally explicit. It found that evidence concerning autism and the overall immunization schedule was limited in both quantity and quality. The report discussed questions involving timing, spacing, same-day combinations, ordering, interactions with health conditions, and cumulative measures of vaccine components. It also recommended research methods capable of examining potentially susceptible groups. If the whole question had already been answered, those recommendations would have been incoherent.

The final word from the IOM is that susceptible subgroups should be sought and studied. That has never happened.

The CDC’s current autism page now states that studies have not ruled out the possibility that infant vaccines contribute to autism and calls for study of mechanisms and potential causal links. That page has been politically contested, and an agency webpage is not scientific proof. It does, however, make Shapiro’s class-wide statement especially difficult to defend. Even the agency at the center of the dispute now distinguishes the evidence concerning MMR and thimerosal from the incompletely studied question of infant vaccines as a whole.

Association is not causation, but a null association is not an all-purpose acquittal

Shapiro’s reasoning depends on an asymmetry that appears constantly in vaccine-safety rhetoric. When a study reports a positive association, we are correctly reminded that association does not prove causation. Confounding, selection, measurement error, reverse causation, multiple testing, and chance can produce a signal. When a study reports no association, however, the same people often promote that result as proof that causation is impossible, especially when the studies are analyzed over and over to produce a null result. I am putting my reputation and neck on the line to say plainly: I believe Ben has been duped into citing fraudulent and mis-aimed science.

Even without analysis-to-result (which we know has been used over and over), let me explain how causal inference works. A positive association is evidence whose causal weight depends on design, timing, consistency, dose response, bias, biological coherence, and replication. A null association is evidence that the study did not detect an effect under its definitions, measurements, population, model, and statistical resolution. Strong, repeated null results can make a proposed effect increasingly unlikely, and the MMR literature has done exactly that, but only for a large average effect. Yet no observational study can convert “not detected here” into “cannot occur in any child or subgroup of susceptible children through any pathway.”

This matters because nearly all of the evidence Shapiro invokes is observational association research. Randomizing children to the full recommended schedule, an alternative schedule, or no vaccination for years of autism follow-up would create ethical and practical problems, and the Institute of Medicine advised against such a trial. Observational studies are therefore indispensable. Shapiro cannot dismiss positive observational findings because they are “only associations” while treating negative observational findings as causal absolution. The standard has to be the same in both directions.

Association does not test cauasility. Again, the wrong type of study.

His severity argument does not test vaccination

Shapiro tried to explain rising autism prevalence by pointing to a 2026 paper on adaptive functioning. The paper found that diagnosed autism with mild, borderline, or no significant adaptive challenges increased between 2000 and 2016, while the estimate for moderate-to-profound adaptive challenges declined slightly from 1.5 to 1.2 per 1,000 children. The authors wrote that the trend could reflect improved identification of milder autism and possibly greater access to therapies that improved functioning. Their paper is useful for understanding changes in the population being identified. It contains no analysis of vaccine exposure.

It therefore cannot carry the causal weight Shapiro placed on it. Diagnostic expansion and a real increase are not mutually exclusive. Changes in ascertainment can operate alongside changes in environmental risk. But were doctors in the 1950s, 1960s, 1970s, and 1980s so bad they could not see that a child could not speak at age 3, 4, 5, 6, 7, and 8? Pediatricians do not do diagnosis anyway: speech language pathologists do. Adaptive functioning can also improve with treatment, making severity at age eight an imperfect proxy for severity at onset or for the number of children who experienced developmental regression. A different CDC-network analysis, using a different definition of profound autism, estimated that profound-autism prevalence rose from 2.68 per 1,000 in 2000 to 4.59 per 1,000 in 2016. That does not automatically invalidate the adaptive-functioning study; it shows why definitions and denominators must be examined before a commentator turns one severity trend into a complete explanation of autism prevalence.

Shapiro compounded the problem with an avoidable factual error. He dated the Americans with Disabilities Act to 1975. The ADA was enacted in 1990. Congress passed the Education for All Handicapped Children Act in 1975; the law was renamed the Individuals with Disabilities Education Act in 1990, when autism and traumatic brain injury were added as disability categories. Unless he had an unstated reason for that reference, Shapiro may have blended those events together, but a thesis built around legislative incentives ought to get the legislation right if not.

Wakefield deserved investigation from both sides

Shapiro’s Wakefield account contains important truths. The 1998 Lancet report was a case series of twelve children. Parent-reported timing could generate a hypothesis but could not establish causation. The article itself did not prove an MMR-autism association, and it is true that The Lancet later retracted it. Large subsequent studies did not support a broad MMR-autism relationship. Anyone presenting the case series as decisive causal proof would be overstating it. My friend and colleague, Dr. Andrew Wakefield would agree.

Wakefield’s hypothesis was that something was going on with GI issues following MMR vaccine in children with autism. If Ben had read the study, he would have seen that the pilot study in fact, it never claimed, or even proposed that vaccines might be involved as a causal factor in autism.

Shapiro nevertheless repeated the institutional case as though no adversarial record existed. Before presenting Wakefield as the origin and embodiment of the entire question, he should have examined the chronology of the litigation funding, the dispute over whether the children’s procedures were clinical investigations or research, the referral histories, the patent claims, the General Medical Council record, and the defense offered by Wakefield and his supporters. Mary Holland’s account argues that the litigation grant concerned a later project, that the Lancet children were investigated for clinical reasons, that referrals did not occur as the prosecution alleged, and that the relevant patent was held by the hospital and concerned a therapeutic product rather than a competing preventive measles vaccine. Those are defense claims, not established facts, and they should be checked against the underlying documents. Due diligence means reading them before rejecting them.

Ben should know that it was clear at the time that the reason why Wakefield was Wakefield was the fear that parents might refuse vaccines for their child if it was acknowledged that vaccines might cause autism.

Is that reason? Logic? Does that lead to knowledge? Or does the denialist agenda lead to an autism epidemic, altering the lives of millions of children and their families?

The 2012 High Court judgment in the appeal of Wakefield’s coauthor John Walker-Smith makes the duty to pay attention to the reasoning used unavoidable. The court found that the General Medical Council panel’s conclusion rested on “inadequate and superficial reasoning” and, in a number of instances, wrong conclusions. It quashed both the finding of serious professional misconduct and the sanction of erasure. The judgment does not exonerate Wakefield, who did not pursue the same appeal, and it does not validate an MMR-autism hypothesis. It does establish that a material part of the disciplinary narrative failed under judicial scrutiny.

I am not asking Shapiro to believe Wakefield. I am asking him to do what he demands of journalists examining Anthony Fauci: read the adverse record, separate proved findings from accusations, and stop treating institutional punishment as a substitute for evaluating evidence.

The William Thompson matter cannot honestly be erased

Shapiro also ignored the most serious unresolved research-integrity controversy involving a CDC MMR study. William Thompson was a coauthor of the CDC’s 2004 Pediatrics paper on age at first MMR vaccination and autism. In a 2014 statement issued through counsel, Thompson said that the authors had “omitted statistically significant information,” that the omitted analysis suggested increased risk among African American males vaccinated before 36 months, and that he believed the final study protocol had not been followed. In the same statement, he said vaccines save lives, advised no parent to avoid vaccination, and judged vaccine benefits to outweigh the risks. His position was not the cartoon Shapiro’s narrative requires.

The Congressional Record contains a longer account attributed to Thompson. It says the team had finalized an analysis plan in advance, that a race effect appeared in the sample not restricted to children with Georgia birth certificates, that the authors decided not to report race effects, and that controversial findings were intentionally withheld. In recorded conversations with Brian Hooker, Thompson said he had run analyses for the group over several years and met with the coauthors weekly to discuss results. He described the restriction of the race analysis to the birth-certificate subgroup as the study’s biggest problem and said the team had deviated from what it had agreed to do. The transcript is publicly available.

CDC gave a substantive response. It said birth certificates supplied more reliable race information and covariates such as birth weight, maternal age, and education, which were unavailable for children without those certificates. CDC also argued that the association observed for vaccination between 24 and 36 months likely reflected immunization requirements for enrollment in preschool special-education programs. That response belongs in the record, just as Thompson’s objections do.

Hooker’s numerical reanalysis cannot settle the dispute. The journal retracted it after concerns about the statistical analysis, the validity of its conclusions, and undisclosed conflicts. That retraction weakens Hooker’s reported effect estimate and makes it unsuitable as proof of causation. It does not erase Thompson’s signed statement, the protocol question, or the existence of multiple dated analyses.

I have previously reported more detailed allegations: that Coleen Boyle directed changes to age-group definitions after the signal was known, that the valid-birth-certificate restriction removed many valid data points and reduced power, that the African American boys subgroup analysis was omitted by Frank DeStefano from his presentation to the Institute of Medicine, and that Thompson was placed on leave during the relevant period to prevent him from reporting the result after approaching then-CDC Director Judy Gerberding with the results. Those allegations have not been independently adjudicated in a process that released and tested the complete documentary record. But they come from Thompson himself. The answer is in the record straightforward: we have the original protocol, dated analysis files, code, correspondence, presentation drafts, final slides, and complete data and independent reconstruction corroborates. Shapiro did not examine the accusation, the defense, or the material needed to resolve it. He simply acted as if these events never happened.

“No data” is factually indefensible

Shapiro could have argued that studies reporting positive vaccine-autism associations are methodologically weak. Many are. A 2013 paper by Geier and colleagues reported an association between thimerosal-containing vaccine exposure and autism diagnoses, but its use of passive reporting and administrative data creates serious vulnerability to reporting bias, selection, and confounding. A 2014 study by Deisher and colleagues reported ecological correlations between autism-prevalence change points and vaccines manufactured with human fetal cell lines. Ecological timing cannot establish individual-level causation. Gallagher and Goodman reported higher odds of parent-reported autism among boys vaccinated against hepatitis B in the neonatal period, but that paper relied on a very small number of autism cases and was retracted in May 2026 after an independent statistical review identified major defects.

The proper conclusion is that these papers do not prove that vaccines cause autism. The improper conclusion is that they do not exist. WHO’s own 2025 review identified eleven studies reporting potential associations and then judged their evidentiary strength very low. That is an appraisal. “There is no data, none” is an erasure.

Weak positive studies can still perform a useful scientific function when they identify testable exposure windows, products, or subgroups for better research. Some will fail replication. Some signals will disappear when confounding is controlled. Others may survive. The point of science is to find out which is which. Declaring an empty record before reading the adverse evidence is not skepticism; it is message discipline.

The funding gap did not arise by accident

Shapiro speaks as though the quantity of research reflects only the inherent merit of the question. Research agendas are also shaped by prior institutional judgments. In 2009, the Interagency Autism Coordinating Committee declined to approve two objectives that specifically called for additional vaccine research. The Autism Science Foundation celebrated the decision, while noting that vaccines could still be examined under broader environmental and susceptible-population initiatives. Its public explanation was candid: the committee believed vaccine research should no longer be called out because the question had already been resolved. The decision is part of the public record.

This creates a closed loop. Institutions declare the question settled, targeted funding contracts, the resulting scarcity of product-specific and subgroup-specific studies is then cited as evidence that no serious question remains, and commentators such as Shapiro announce that there are no data. A funding decision does not prove a suppressed causal relationship, but it can explain why the exact studies needed to test a narrowed hypothesis were never prioritized. Absence of funded investigation is not the same thing as a negative result.

He did not look for a mechanism

Shapiro gives no indication that he searched for a serious mechanistic model. He could have found my 2016 book, The Environmental and Genetic Causes of Autism. He could have found my freely available 2018 review, “Autism Is an Acquired Cellular Detoxification Deficiency Syndrome with Heterogeneous Genetic Predisposition”, or the accompanying IPAK ASD causality model.

The model integrates heterogeneous genetic susceptibility with toxicant exposure, endoplasmic-reticulum stress, the unfolded-protein response, impaired mitochondrial function, disrupted cellular detoxification, cytokine signaling, neuroinflammation, chronic microglial activation (immunoneuroexcitotoxicity sensu Blaylock), and altered neuronal and glial survival leading to altered neurodevelopment. It proposes that different families can arrive at overlapping behavioral diagnoses through different combinations of susceptibility and exposure. It also generates testable predictions about developmental timing, gene-by-environment interaction, metabolic state, dose, recovery capacity, and biological markers.

The model is a hypothesis with many, many lines of corroborating evidence. It is not proof that any vaccine causes autism, and it should not be sold as such because science does not use the currency of proof. Its links must be tested in human studies with accurate exposure histories, relevant biomarkers, prespecified interaction terms, and independent replication. If we can prevent autism by finding who might be at highest risk due to vaccine effect, causality is moot.

Shapiro was entitled to challenge the model, identify an unsupported step, show that its predictions failed, or argue that another model better explains the observations. He did none of those things because he appears not to have read the paper, the model, or my book. He should call for the right studies, not rely on the absence of evidence.

The biological observations underlying this line of inquiry are not imaginary. Postmortem work has reported microglial activation and increased microglial density in autistic brain tissue. Other work found an active neuroinflammatory process involving microglia, astroglia, and cytokines, and PET imaging reported excessive microglial activation in multiple brain regions. These findings do not identify the initiating exposure. They do show that chronic neuroimmune activation is a legitimate part of autism biology and that mechanistic questions involving immune activation cannot be dismissed merely because an average epidemiological estimate is null.

What an evidence-based conclusion would sound like

A responsible review would say that large, well-designed studies strongly weigh against MMR as a major cause of autism across the general population. It would say that the evidence against a population-wide association is much stronger than the positive evidence offered by ecological studies, passive-reporting analyses, small cross-sectional studies, or retracted reanalyses. It would also say that not every vaccine and schedule-level hypothesis has been examined with equal quality, that some Institute of Medicine conclusions remain “inadequate to accept or reject,” that rare susceptible subgroups are difficult to define and study, and that the Thompson controversy warrants documentary resolution rather than ritual dismissal.

The next studies should be independent, preregistered, product-specific, and adequately powered for biologically defined interaction effects. They should use individual vaccination dates and formulations, co-administer vaccines and validated developmental phenotypes, clinically relevant exposure windows, sibling and within-family comparisons where appropriate, negative controls, active safety surveillance, and transparent code. Schedule research should examine cumulative and same-day exposure without pretending that antigen count is a complete proxy for every immunological or toxicological property. Any positive subgroup finding should be replicated before clinical claims are made. Any null finding should report confidence intervals capable of showing which risk magnitudes were actually excluded.

I would change my mind if repeated, independent studies used those methods and consistently excluded meaningful product-, timing-, and susceptibility-specific effects while the mechanistic predictions also failed. I would change my mind if, using family history, genetics, immune markers, and other lines of clinical and demographic evidence we failed to be able to prevent autism in the unvaccinated-at-risk compared to the vaccine-at-risk.

We need to move beyond association to prevention, and I invite skeptics and those interested to come to ipakedu.online where I will make my white paper, Beyond Association, available to members. That said, evidence favoring causation would require the opposite pattern: replicated associations in prespecified susceptible groups, temporality, dose or timing response, convergence across designs, and human mechanistic findings that connect exposure to the relevant developmental pathology. That is how a disputed hypothesis should be narrowed or rejected.

Shapiro offered none of this. He converted “the best evidence argues against a large average MMR effect” into “there is no data linking any vaccine to autism.” He converted a study of diagnostic severity into a vaccine-safety study without a vaccine variable. He repeated one side of the Wakefield case, ignored the Walker-Smith judgment, omitted William Thompson, overlooked positive studies instead of appraising their weaknesses, and never confronted the funding conditions that helped determine which questions were studied. He did not engage the mechanistic literature because he did not appear to know it existed.

The irony is hard to miss. Shapiro ended by blaming Anthony Fauci for destroying public faith in data. Yet Shapiro committed the same institutional offense in miniature: he began with the approved conclusion, selected the portion of the record that supported it, and treated confidence as a substitute for completeness. Data do not lie, but they do not answer questions they were never designed to test, and they cannot rescue a commentator who reads the verdict without reading the record.

Share

Leave a comment

Share

Subscribe now

James Lyons-Weiler, PhD
Popular Rationalism
August 18, 2026

On August 17, 2026, the U.S. Department of War ordered 30 American academic institutions to conduct immediate audits of their academic, financial, and research relationships with foreign entities the government considers security risks. The reviews must examine collaborations involving institutions on the Department’s Section 1286 list and organizations associated with rebranded Confucius Institutes. Universities were directed to determine whether sensitive or export-controlled research has been exposed, mitigate identified risks, terminate problematic relationships where necessary, and report their findings by August 31 if they wish to maintain eligibility for future federal research funding.

The Department’s public announcement does not identify the 30 institutions. A U.S. official provided the list to DefenseScoop, however, and subsequent reporting identifies Harvard, MIT, Duke, Georgetown, NYU, Johns Hopkins and other major research universities among those notified. Harvard therefore appears on the audit list according to reporting based on a government source, rather than from a publicly released Department list. That distinction should be preserved.

Four days earlier, two House committees had released a 53-page investigation of Harvard titled Compromised Independence: CCP Influence at Harvard University. Among its findings was an allegation that deserves considerably more attention than the generalized accusation that Harvard has “ties to China.” The committees identified a Harvard-affiliated nonprofit, Harvard Global Research & Support Services, Inc., through which foreign-sponsored research can be formally awarded even while Harvard University personnel and resources conduct the work. The committees allege that the arrangement “appears to be designed to circumvent” federal foreign-funding disclosure requirements.

That allegation has not yet been proved. The structure, however, is real.

And after forty years of sporadic enforcement of a federal foreign-funding disclosure law, the structure gives investigators a remarkably clean question to answer.

The Question Is Whether the Money Remained Reportable

Section 117 of the Higher Education Act is fundamentally a transparency statute. An institution receiving federal financial assistance generally must report gifts from and contracts with a foreign source when the value reaches $250,000 in a calendar year, whether through one transaction or aggregated transactions with the same source. The law does not prohibit universities from taking foreign money. It requires qualifying foreign money to be disclosed.

That distinction matters because the Harvard Global question is not resolved merely by showing that Harvard created a separate nonprofit.

Harvard Global’s own current website says that, depending on sponsor requirements, Harvard University may ask Harvard Global to serve as the proposal applicant and awardee. Harvard Global then formally accepts the sponsored funding, while “Harvard University researchers perform all research activities using University resources” under contract to Harvard Global. The organization describes its status as “separate-but-affiliated” and says that this allows it to support sponsored activities that Harvard University cannot support because of sponsor requirements or legal limitations overseas.

Harvard Global also says that all projects proposed through and awarded to it are recorded in Harvard’s Grants Management Application System, that University Office for Sponsored Programs personnel also hold appointments with Harvard Global, and that Harvard schools and departments retain responsibilities for approvals, research oversight, facilities and other support. U.S.-based Harvard researchers working on these awards remain on the Harvard University payroll. Harvard Global expressly says that U.S. federal sponsored awards do not flow through its entities.

Those facts are relevant counterevidence to any claim that Harvard Global necessarily operates as a secret off-the-books organization. Its current public description portrays an integrated and internally recorded sponsored-research mechanism.

But internal visibility is not the same question as federal disclosure.

The House committees reproduced Harvard Global’s 2011 IRS Form 990, which stated that the organization would “seek research grants from foreign grantmakers” and would then use those funds to pay Harvard faculty, students and others to perform the research. The committees also reproduced an earlier version of Harvard Global’s website saying that when Harvard University could not accept a sponsor’s terms for legal or financial reasons, Harvard Global could act as applicant and awardee while Harvard researchers performed the work using University resources.

The exact wording highlighted by the committees has since changed. The functional arrangement has not disappeared from the current site. Harvard Global still says it may formally receive an award while Harvard researchers conduct the research using University resources, and that its separate legal status permits sponsored activities Harvard University itself cannot support under certain sponsor or overseas legal requirements.

That is precisely where Section 117 becomes relevant.

The Department of Education’s November 2023 Section 117 guidance specifically addresses intermediaries. Education states that a university receiving the benefit of a foreign gift or contract may still have to report it when the benefit passes through an intermediary. The Department warns that allowing institutions and foreign sources to avoid reporting merely by inserting another entity into the transaction would defeat the statute. It places the ultimate responsibility for reasonable due diligence on the institution receiving the benefit.

The guidance itself also contains an important legal restraint: Education expressly says its FAQ does not create new law and that intermediary questions require individualized, fact-specific assessment.

So the evidentiary burden is straightforward.

For each qualifying foreign-origin award received through Harvard Global, identify the ultimate source, value, contractual terms, Harvard personnel and resources benefiting from the award, and the corresponding Section 117 disclosure.

If the transactions appear in the federal reporting record, the circumvention allegation weakens substantially.

If qualifying transactions benefited Harvard but disappeared from federal disclosure because the immediate recipient was Harvard Global rather than Harvard University, investigators will have something much more serious than an unusual organizational chart.

They will have transactions.

Congress Has Been Waiting for This Ledger Since 1986

The foreign-funding problem did not begin with China, Harvard or the Trump administration.

Congress first required American colleges and universities to publicly disclose certain foreign gifts and contracts in 1986. The Department of Education’s 2020 review of the statute’s legislative history says the original concern involved foreign money potentially affecting academic integrity. According to that review, donations from Arab countries associated with Georgetown University’s Center for Contemporary Arab Studies helped motivate the legislation; Japanese corporate relationships with American research institutions were among the other concerns of the period.

This history matters because Section 117 was not written as an anti-China statute. Its underlying principle was broader: when federally supported universities receive substantial financial support from foreign sources, the public should be able to see the relationship.

The enforcement history was much less impressive than the principle.

A bipartisan Senate Permanent Subcommittee on Investigations report in 2019 examined Confucius Institutes operating at American universities. The subcommittee determined that China had provided more than $158 million directly to U.S. schools for Confucius Institutes since 2006. Nearly 70 percent of schools that had received more than $250,000 from Hanban, the Chinese government entity then administering the program, had failed to report that funding properly to the Department of Education. The Senate report also noted that Education had last issued Section 117 guidance in 2004, the same year the first U.S. Confucius Institute opened, and then remained largely silent as the program expanded to more than 100 institutions.

That was an enforcement failure hiding in plain sight.

The federal government then began paying serious attention. Education’s Office of General Counsel initiated a series of Section 117 investigations beginning in 2019. By February 2026, Education reported that American universities had disclosed $67.6 billion in foreign funding since Section 117 was enacted in 1986 and, remarkably, that the majority of that money had been disclosed only since 2019. More than $2 billion in reportable gifts and contracts were filed late during a period in 2025 alone.

A disclosure law can exist for decades while the information system beneath it remains badly incomplete.

The Department of Education’s own inspector general demonstrated how incomplete the enforcement architecture still was as recently as February 2025. The OIG found that Federal Student Aid’s Section 117 oversight consisted largely of reviewing whistleblower tips, Department news releases and media reports and providing technical assistance. FSA had no monitoring plan, policies or procedures for overseeing compliance. The OIG recommended a monitoring system, annual certifications by senior university officials and stronger controls in the reporting portal.

That finding should prevent the current controversy from being reduced to a morality play about one university.

Universities had disclosure obligations. The federal government had enforcement responsibilities. Both matter.

When an oversight system depends heavily on institutions telling the government what the government does not independently know to ask for, sophisticated institutions will inevitably become the stress test. Affiliates, foundations, overseas campuses, research institutes, foreign subsidiaries, subawards, joint programs and intermediary organizations are precisely where a simplistic “Who received the check?” model begins to fail.

Section 117 eventually had to learn to follow the benefit of a transaction rather than merely the name printed beside “payee.”

Foreign Funding Became a Research-Security Problem

Another change occurred during the same period. Federal concern moved beyond the question of who funded the university toward a more difficult question: what research relationships did the funding and affiliations create?

The Charles Lieber case provides one important Harvard example, but it must be described accurately. Lieber, the former chair of Harvard’s Department of Chemistry and Chemical Biology, was convicted in 2021 of making false statements to federal authorities, filing false tax returns and failing to report a foreign bank account in connection with his relationship with Wuhan University of Technology and China’s Thousand Talents Program. His Harvard laboratory had received more than $15 million in NIH and Defense Department research support.

That case is sometimes invoked as though it proved Harvard itself knowingly participated in the arrangement. The Justice Department said the opposite on that particular point: Lieber’s Wuhan affiliation and Thousand Talents participation were “unbeknownst to his employer, Harvard University.”

The distinction is useful. An individual researcher’s concealment cannot simply be transferred into an allegation of institutional intent.

But the case demonstrated why research sponsors could no longer treat undisclosed foreign appointments, laboratories, support and commitments as ordinary bookkeeping mistakes. NIH had already begun addressing foreign-interference concerns by 2018, including undisclosed foreign support and affiliations, intellectual-property risks and breaches of peer-review confidentiality.

Federal policy then became progressively more systematic.

National Security Presidential Memorandum 33, issued in 2021, established a government-wide research-security framework. The CHIPS and Science Act of 2022 added statutory research-security requirements, including provisions concerning malign foreign talent recruitment programs. NSF subsequently implemented certifications and research-security training requirements for senior personnel and institutions.

In 2023, the Defense Department adopted risk-based security reviews for fundamental research selected for funding. The policy requires assessment of potential conflicts of interest and conflicts of commitment arising from foreign influence and allows mitigation when a relationship presents a research-security risk. The policy also makes an important point that should survive the political temperature surrounding this subject: international scientific collaboration is valuable, and risk reviews are supposed to operate without indiscriminately discouraging it.

That is the correct standard.

Foreign collaboration is not evidence of wrongdoing. Chinese co-authorship is not evidence of espionage. Dual-use potential is not proof that technology was transferred unlawfully. A university appearing on a federal audit list is not a finding of guilt.

But those propositions do not eliminate the obligation to examine a collaboration when federal research funding, military-linked institutions, sensitive technologies, undisclosed affiliations or export controls intersect.

The House Harvard report illustrates the difference. Its investigators say they identified more than 140 publications containing both a Harvard-affiliated author and an author affiliated with one of China’s “Seven Sons of National Defense” universities. The committee’s analysis found engineering to be the dominant subject area.

That count establishes co-authorship. It does not by itself establish illicit technology transfer.

The next analysis has to be paper-specific and project-specific. Which collaborations involved U.S. government funding? Which involved institutions subject to federal restrictions at the time the work occurred? What technology, data, equipment or know-how crossed institutional boundaries? What did Harvard’s compliance system review? What did federal sponsors know? Was the work fundamental research intended for open publication, controlled technology subject to restrictions, or something between those categories?

Those questions are much harder than counting publications.

They are also the questions a serious research-security system is supposed to answer.

The Department of War’s Section 1286 program now formally identifies foreign academic and research institutions considered to present elevated risks of misappropriation of U.S.-funded research. Its FY2025 list, released in July 2026, contains 130 institutions in China, Russia and Iran. Less than a month later came the order requiring 30 American institutions to audit their active relationships against that security architecture.

That is a different system from Section 117, but the two systems are beginning to converge.

One follows foreign money. The other follows research exposure. A competent government should be able to connect them.

Harvard’s Disclosure Record Was Already Unresolved

Harvard was already under Section 117 scrutiny before the House released its Harvard Global finding.

Education opened a Section 117 compliance review of Harvard in 2020. That investigation was closed in December 2024 after Harvard entered an attestation and agreement stating, among other things, that it was current in its Section 117 submissions, had written procedures and systems designed to achieve compliance, and had not knowingly or willfully failed to comply.

Four months later, Education opened another review.

On April 18, 2025, the Department announced that its examination of Harvard’s recent disclosures had identified “incomplete and inaccurate disclosures.” The underlying notice referred to noncompliant transaction disclosures, anonymous donors and deficiencies that also appeared in amended reporting covering 2014 through 2019. That investigation was still open when the House committees issued their August 2026 report.

Those findings do not prove the Harvard Global allegation either. They do make it impossible to dismiss the intermediary issue as a purely hypothetical concern.

Education’s February 2026 disclosure release adds scale. It reports that Harvard disclosed more than $610 million from counterparties located in statutory “countries of concern” between 1986 and December 16, 2025, the highest amount among the institutions identified by Education under that particular measure. That figure should not be casually relabeled “Chinese funding”; Education’s category includes multiple countries of concern. The House report separately cites Education data for its own claim concerning Harvard’s China-attributed funding. They are related measures, not interchangeable ones.

This is exactly where careless reporting can ruin a legitimate investigation. A large foreign-funding number does not prove influence. A Chinese coauthor does not prove military transfer. A nonprofit intermediary does not prove evasion.

The case becomes strong by refusing to transfer evidence from one proposition to another.

What is established is that Harvard Global is a separate Harvard-affiliated nonprofit capable of receiving foreign-sponsored awards; Harvard researchers and resources can perform the resulting work; Education says intermediary transactions benefiting a university can remain reportable; Harvard has experienced repeated Section 117 scrutiny; and the current federal investigation has identified problems in Harvard’s disclosure submissions.

What has not yet been publicly established is the transaction-level proposition necessary to prove circumvention: that Harvard Global received qualifying foreign-origin gifts or contracts benefiting Harvard and that Harvard failed to make Section 117 disclosures that the law required.

That ledger is now the evidence that matters.

Audit the Transactions, Not the Narrative

The federal government has spent forty years building toward a remarkably ordinary accounting exercise.

Take every Harvard Global foreign-sponsored award within the relevant period. Identify the true foreign source and any intermediary. Record the amount and date. Identify the Harvard school, principal investigator and personnel performing the work. Identify the University facilities, equipment and administrative resources used. Determine whether the transaction crossed the Section 117 threshold alone or when aggregated with other transactions from the same source. Then locate the disclosure Harvard filed with Education.

For research-security review, extend the same record to foreign collaborators and institutions. Determine whether any collaborator was on the applicable Section 1286, Entity List or other relevant federal restriction list at the time. Determine the source of U.S. government funding, the nature of the research, applicable export-control status and the mitigation or approval record.

This is what an audit is for.

It should also be done across the other 29 institutions. The history makes clear that foreign-funding underreporting and inadequate federal oversight were never exclusively Harvard problems. A 2019 bipartisan Senate investigation found widespread reporting failures involving Confucius Institute funding. Education’s inspector general found an inadequate federal monitoring system in 2025. Education now says that the majority of the $67.6 billion disclosed since the law was enacted in 1986 was reported only after 2019.

Harvard deserves neither immunity because of its prestige nor a presumption of guilt because it is Harvard.

The same evidentiary rule should apply to every institution.

The House committees have produced enough evidence to justify examining Harvard Global closely. Harvard Global’s own documents establish the intermediary mechanism. Education’s own guidance establishes that an intermediary does not necessarily sever the reporting obligation. Harvard’s recent Section 117 history establishes that its disclosure system has already produced findings serious enough for Education to reopen an investigation. The War Department now has an independent research-security reason to examine foreign collaborations at Harvard and dozens of other institutions.

The remaining questions are factual.

Which foreign awards passed through Harvard Global?

Which of those awards conferred a reportable benefit on Harvard University?

Where are those transactions in Harvard’s Section 117 disclosures?

Did Education examine Harvard Global when it closed the first Harvard investigation in December 2024?

Are any Harvard Global transactions among the incomplete or inaccurate disclosures Education identified four months later?

For the Harvard collaborations involving military-linked Chinese universities, which projects actually involved federal funding, controlled technology, restricted entities or other specific research-security risks?

Those questions can be answered with records.

If the records reconcile, investigators should say so.

If they do not, the government should identify the missing transactions, the responsible reporting entity, the applicable legal requirement and the consequence of the omission.

After forty years, another argument about whether American universities have “foreign ties” would tell us very little.

The government finally has the tools to ask the narrower and more useful question:

What happened to the money and the research after they crossed the institutional boundary?

That is where the record begins.

Source Notes

1. U.S. Department of War (2026). “Department of War Orders Research Security Audits at 30 Academic Institutions,” August 17, 2026. The order requires the notified institutions to review academic, financial and research collaborations involving Section 1286 entities and organizations associated with rebranded Confucius Institutes and report by August 31.

2. U.S. House Committee on Education and the Workforce and House Select Committee on the Chinese Communist Party (2026). Compromised Independence: CCP Influence at Harvard University, August 13, 2026. The report contains the Harvard Global finding, reproduces prior Harvard Global website language and the nonprofit’s early Form 990, discusses Harvard’s Section 117 history, and reports the committees’ analysis of Harvard collaborations with “Seven Sons” universities.

3. Harvard Global Research & Support Services, Inc. (2026). “International Grant Administration.” Current Harvard Global materials describe its separate-but-affiliated structure, sponsored-award process, use of Harvard University researchers and resources, GMAS documentation and foreign sponsors.

4. U.S. Department of Education, Federal Student Aid (2023; updated 2026). “Frequently Asked Questions: Section 117 Foreign Gift and Contract Reporting.” The guidance explains the Department’s treatment of intermediaries, reasonable due diligence and the institution’s ultimate compliance responsibility while expressly stating that the FAQ does not itself create new legal standards.

5. 20 U.S.C. § 1011f. Section 117 of the Higher Education Act, governing disclosure of qualifying foreign gifts and contracts by federally assisted institutions of higher education.

6. U.S. Department of Education, Office of General Counsel (2020). Institutional Compliance with Section 117 of the Higher Education Act of 1965. The report reviews the 1986 legislative history and the Department’s renewed Section 117 enforcement effort.

7. U.S. Senate Permanent Subcommittee on Investigations (2019). China’s Impact on the U.S. Education System. The bipartisan staff investigation reported more than $158 million in direct Chinese funding for Confucius Institutes since 2006 and found substantial Section 117 underreporting among institutions exceeding the statutory threshold.

8. U.S. Department of Education, Office of Inspector General (2025). FSA’s Oversight of Section 117 Reporting Requirements, February 12, 2025. The OIG found that FSA lacked monitoring plans, policies and procedures for Section 117 oversight and issued six recommendations.

9. U.S. Department of Education (2025). “U.S. Department of Education Initiates Records Request from Harvard University After Discovering Inaccurate Foreign Financial Disclosures,” April 18, 2025, together with the underlying notice of investigation.

10. U.S. Department of Education (2026). “U.S. Department of Education Releases Latest Foreign Funding Disclosures from Federally-Funded American Universities,” February 11, 2026. Education reported more than 8,300 transactions worth over $5.2 billion for 2025, $67.6 billion disclosed since 1986, and stated that the majority of the historical total had been disclosed since 2019.

11. U.S. Department of Justice (2021, 2023). Charles Lieber conviction and sentencing materials. The records establish Lieber’s false statements and undisclosed Wuhan University of Technology and Thousand Talents relationships while also stating that those relationships were unbeknownst to Harvard.

12. U.S. Department of Defense (2023). “Department of Defense Strengthening Efforts to Counter Unwanted Foreign Influence in Department-Funded Research at Institutions of Higher Education” and Policy for Risk-Based Security Reviews of Fundamental Research.

13. National Science Foundation (2022–2026). CHIPS and Science Act research-security implementation materials, including research-security training and malign foreign talent recruitment program certification requirements.

14. U.S. Department of War (2026). Updated Section 1286 foreign-institution list, identifying 130 academic and research institutions in China, Russia and Iran considered to present elevated risks of misappropriation of U.S.-funded research and development.

But that is not what AAI did.

Instead, in its August 17 article, AAI told its readers that the executive order contains “calls to replace aluminum-based adjuvants,” characterized those calls as unsupported by science, and answered them with the observation that aluminum-containing adjuvants “have been used safely in vaccines for decades.” AAI then shifted immediately to autism, despite the fact that Executive Order 14420 does not mention autism.

The underlying executive order says something materially different. It directs HHS to “develop additional alternative adjuvants to aluminum and conduct comparative safety and efficacy studies.” It does not order aluminum adjuvants removed. It does not declare them unsafe. It directs the government to develop alternatives and compare them. The same section requires continuing risk-benefit evaluation, stronger vaccine-safety monitoring, and research.

For an association of immunologists to characterize comparative adjuvant research as a species of vaccine misinformation is difficult to reconcile with the experimental literature of immunology itself.

This is not a semantic objection. Immunologists know what an adjuvant is. They know why aluminum hydroxide is placed next to an antigen. They know that it is not there as a preservative, a diluent, or a manufacturing remnant. It is deliberately included because it changes what the immune system does with antigen. It induces tissue injury and danger signaling, recruits innate immune cells, changes antigen uptake and presentation, promotes germinal-center responses, and strongly favors humoral immunity, including IgG1 and IgE under appropriate conditions. Your own field has spent decades exploiting precisely those properties.

That is why the animal literature AAI omitted matters.

Researchers use aluminum hydroxide in experimental protocols to produce pathological immune phenotypes so that therapies for those phenotypes can be studied. They use it to establish allergic asthma. They use it to establish allergic rhinitis. They use it in peanut-allergy and anaphylaxis models. They use it with prostate autoantigens to establish experimental autoimmune prostatitis. In genetically susceptible mice, repeated alum administration without a deliberately supplied foreign antigen has induced a Sjögren’s syndrome-like disorder. In another elegant experiment, aluminum salts generated a tissue environment in which an antigen encountered days later—not physically mixed with the alum—elicited a strong IgE response.

These findings do not prove that every aluminum-adjuvanted vaccine causes these diseases in humans. That is not the claim, and immunologists should not need that elementary distinction explained to them.

They establish something that is sufficient to demolish AAI’s argument: there is a substantial experimental basis for asking whether the immunologic effects of aluminum adjuvants vary with antigenic context, dose, timing, repetition, host susceptibility, and persistence, and whether alternative adjuvants can achieve the desired immunity with a different risk profile.

That is a scientific question. The executive order asks for it to be studied.

AAI calls asking it “misinformation”.

Subscribe now

Read the animal protocols, not the public-relations summary

Consider experimental autoimmune prostatitis.

Zhang and colleagues did not report an uncontrolled observation in a mouse that happened to receive aluminum. They designed a controlled experiment. Forty C57BL/6 mice were divided among four groups: T2 prostate peptide plus aluminum hydroxide, T2 peptide alone, aluminum hydroxide alone, and saline. The prostate lesions were most severe in the mice receiving T2 plus aluminum hydroxide. TNF-α and CRP increased, prostate IL-1β increased, and the authors concluded that injection of T2 peptide in aluminum hydroxide successfully established experimental autoimmune prostatitis. Their paper is titled Establishment of experimental autoimmune prostatitis model by T2 peptide in aluminium hydroxide adjuvant.

Those words deserve more attention than AAI gives them.

The disease model was not already there waiting for aluminum hydroxide to “exacerbate” it. The experimental autoimmune disease was established by the induction protocol. The prostate antigen supplied target specificity; the aluminum hydroxide supplied adjuvant activity. Together they produced the autoimmune inflammatory phenotype the investigators wanted.

There is a useful causal distinction here that is routinely blurred in vaccine-safety discussions. A cause need not be sufficient by itself to be causal. Cigarette smoke need not cause lung cancer in every smoker to be a cause of lung cancer. Nor need it be the sole cause of lung cancer. A particular HLA allele need not independently produce an autoimmune disease to participate causally in the disease. Infection, molecular mimicry, tissue injury, defective regulation, an autoantigen, and an inflammatory stimulus may each be component causes in a sufficient causal constellation.

The same rule applies to an adjuvant.

It makes no immunologic sense to reply that aluminum hydroxide “did not cause the prostatitis by itself.” Of course it did not supply the prostate antigen. That is why the experiment contains an autoantigen. The relevant observation is that T2 peptide alone and aluminum hydroxide alone were experimental controls, while T2 plus aluminum hydroxide successfully established the autoimmune disease model.

That is causation within the experimental model, not merely correlation.

And the authors tell us why such models matter: they are built to investigate pathogenesis and treatment.

We deliberately use aluminum hydroxide to create allergic disease models, then test treatments in them

The allergy literature makes the point even more clearly.

Ji and colleagues published a study whose title should be read slowly by anyone inclined to dismiss research on alternative aluminum adjuvants: AcCystatin, an immunoregulatory molecule from Angiostrongylus cantonensis, ameliorates the asthmatic response in an aluminium hydroxide/ovalbumin-induced rat model of asthma.

The investigators created an OVA/Al(OH)3 asthma group, produced extensive airway inflammatory responses, OVA-specific IgE, eosinophilic inflammation, cytokine alterations, peribronchial and perivascular inflammation, and goblet-cell changes, and then tested whether AcCystatin could prevent or treat the induced disease. It did.

This is not an obscure philosophical inference about what aluminum “might” do.

The investigators needed an asthmatic phenotype in an animal so they could test a possible treatment. Aluminum hydroxide was part of the protocol used to produce it. It was a causal factor.

The same experimental architecture appears in allergic rhinitis. Günel and colleagues established allergic rhinitis in rats through sensitization and challenge with ovalbumin; the published protocol used aluminum hydroxide during sensitization, after which the investigators tested Pycnogenol and corticosteroid treatment against the resulting allergic inflammation. The model produced OVA-specific IgE and the expected inflammatory pathology, which the interventions then reduced.

One can find variations of this design throughout experimental allergy research because aluminum hydroxide is useful for exactly the reason vaccinologists value it: it changes an otherwise inadequate antigen exposure into a more powerful adaptive immune event.

This deserves particular emphasis because AAI cited a 2026 Pediatrics review to reassure its readers about aluminum adjuvants. The basic immunology is not contested even there. Alum is associated experimentally with antigen-specific IgE responses. The disagreement concerns what those mechanisms imply for clinical risk at vaccine doses in human populations—not whether the mechanism exists.

The proper scientific response to that state of knowledge is comparative investigation.

It is not ridicule of comparative investigation.

Subscribe now

Peanut allergy should end any pretense that this question is biologically unserious

The peanut literature is especially instructive because the experimental outcome is not merely a cytokine shift or an antibody titer. Investigators intentionally create IgE-mediated food allergy and anaphylaxis so they can test ways to prevent or treat it.

In one classic experimental design, mouse strains were sensitized to crude peanut protein by intraperitoneal injection with alum. The susceptible strains generated high peanut-specific IgE and IgG1; some subsequently developed anaphylactic symptoms after peanut challenge. Animals given peanut orally without the adjuvant did not develop the same IgE/IgG1 pattern and failed to develop anaphylactic reactions in that experiment.

That comparison is important. It tells us that “exposure to peanut” is not a biologically complete description of the experiment. How antigen is encountered, and in what innate immune context, changes the adaptive result.

Modern investigators continue to use alum-sensitized peanut models specifically to test treatment.

A study developing virus-like-particle vaccines against peanut allergy sensitized BALB/c mice intraperitoneally with peanut extract adsorbed to alum, established the allergic state, and then tested whether a single immunotherapeutic injection could protect those sensitized mice from anaphylaxis.

Another study sensitized mice with peanut extract adsorbed to alum and then tested an anti-Ara h 2 monoclonal antibody; the treatment blocked systemic and local allergic reactions.

Other investigators have developed peanut-anaphylaxis models expressly because animal models are necessary for testing candidate immunotherapies. More recent work has tested pharmacologic inhibition of ORAI1-mediated mast-cell activation against peanut-induced anaphylaxis and acute diarrhea in sensitized mice.

Again, precision matters. Peanut protein supplies the allergen. Aluminum hydroxide does not magically contain Ara h 1 or Ara h 2. The claim is not that injecting aluminum into any mammal invariably produces peanut allergy.

The fact is more relevant to the present policy dispute: aluminum hydroxide is intentionally used as an immunologic causal component of protocols capable of converting antigen exposure into a reproducible pathological IgE-mediated state. Researchers then use the disease they induced to determine whether a proposed therapy can undo it.

Calling comparative investigation of aluminum alternatives “unsupported by scientific evidence” while this literature sits in PubMed is not a defensible position for an immunology society.

Subscribe now

There is also an experiment in which no foreign antigen was deliberately supplied

If AAI’s response is that all of those examples require an antigen, then it should read Bagavant and colleagues.

The investigators used female NZM2758 mice, a genetically susceptible strain that does not spontaneously develop the Sjögren’s-like disease under study. They injected alum and compared those animals with PBS controls. The mice received an initial alum exposure and additional injections at four and eight weeks. Alum-treated animals developed significantly reduced salivary function, increased incidence and severity of sialadenitis, lymphocytic infiltrates, and increased ANA reactivity. The loss of salivary function persisted through the end of the experiment.

The paper’s title is not ambiguous: Alum, an Aluminum Based Adjuvant, Induces Sjögren’s Syndrome-like Disorder in Mice.

The authors concluded that alum induced a chronic Sjögren’s syndrome-like disorder in this genetically susceptible strain and explicitly wrote that the potential of aluminum-based adjuvants to induce autoimmunity should be monitored in genetically susceptible individuals.

There was no deliberately co-administered foreign antigen in that induction protocol.

The obvious objection is that these were genetically susceptible mice.

That objection does not make the result disappear. It identifies the scientific question.

Human beings are not an inbred strain.

They differ in HLA haplotypes, innate immune signaling, regulatory pathways, renal physiology, macrophage biology, autophagy and xenophagy, prior immune history, microbiota, infections, inflammatory state, and background autoantibody repertoires. If an immune-active exposure produces a pathological phenotype preferentially in a susceptible host, the correct translational question is not whether the average host is susceptible. It is whether we can identify the corresponding human susceptibility distribution.

AAI’s appeal to decades of use supplies no answer to that question.

Indeed, animal work on aluminum persistence already tells us that host biology matters. Authier and colleagues established aluminum-hydroxide-adjuvanted vaccine-associated macrophagic myofasciitis lesions in two rat strains and found that lesion persistence differed with genetic background. Their conclusion was that genetic determinants of immune responses could interfere with clearance and condition persistence of the lesions.

That is exactly the kind of heterogeneity a serious safety program should want to characterize.

An adjuvant does not live inside the conceptual boundaries of a vaccine label

There is another assumption in AAI’s reasoning that immunologists should reject immediately: the idea that the only antigen relevant to an adjuvant event is the antigen printed on the vial.

The immune system has never heard of a package insert.

An intramuscular injection enters living tissue containing host proteins, extracellular matrix, resident leukocytes, blood-derived proteins, dying cells, cellular debris, preexisting memory cells, microbial products, and whatever inflammatory biology that particular person brings to the encounter.

Alum itself expands that environment.

Marichal and colleagues demonstrated that aluminum adjuvant causes host-cell death and release of host DNA, and that this endogenous DNA contributes substantially to alum’s adjuvant activity. The released DNA was not merely inert debris; it acted as an endogenous immunostimulatory signal and contributed differently to IgE and IgG1 responses.

McKee and colleagues extended this finding in an intramuscular model particularly relevant to vaccination. Host DNA rapidly associated with injected alum. Degrading extracellular DNA reduced CD4 T-cell priming, while host DNA enhanced MHC class II antigen presentation and prolonged antigen-presenting-cell interactions with CD4 T cells.

That does not prove that a vaccinated child becomes immunologically sensitized to every self protein released at the injection site.

It proves that the closed-vial model is false.

The biologically relevant system contains more than aluminum and the intentionally manufactured antigen. Cell death and endogenous danger signals are part of the mechanism by which alum works. That should lead an immunologist naturally to questions about the antigenic and inflammatory neighborhood in which adjuvant activity occurs.

Then there is the experiment AAI’s argument has even more difficulty accommodating.

Aluminum and antigen do not always have to arrive together

Kuroda and colleagues examined the effects of inhaled fine particulates, including aluminum salts. They administered alum into mouse airways, where it caused alveolar macrophage death and sustained IL-1α release. They then exposed the animals to aerosolized ovalbumin three, seven, or fourteen days later.

The antigen was not premixed with the alum.

It was not given simultaneously.

Yet mice exposed to OVA after alum developed strong OVA-specific IgG1 and IgE responses and airway inflammatory recruitment; the effect persisted even when OVA exposure occurred fourteen days after the particulate exposure. The authors described particulates as producing a long-lasting adjuvant effect on subsequently encountered antigen.

This was an airway experiment, not an intramuscular childhood vaccination study. Route and tissue compartment matter. It would be wrong to pretend that this experiment proves that aluminum injected into a child’s deltoid adjuvants every environmental protein encountered anywhere in the body over the following two weeks.

But that limitation does not reverse the experiment’s meaning.

It establishes a principle that AAI should find extremely interesting: physical co-formulation and simultaneity are not universal requirements for adjuvant-conditioned sensitization. In at least one well-defined tissue system, particulate exposure created a durable innate environment in which a subsequently encountered antigen elicited a type 2 allergic response.

For vaccine safety science, that observation should open questions rather than close them.

How long does the immunologically altered microenvironment persist after intramuscular aluminum administration? How far does it extend? Which antigen-presenting cells participate? What happens during a coincident viral infection? What happens when substantial tissue injury is occurring? What happens in an infant with eczema and impaired barrier function, a child with an evolving food-antigen repertoire, or an individual already carrying autoreactive B-cell clones? What changes after the second, third, or eighth aluminum-adjuvanted exposure?

And how do other adjuvants compare on these questions?

Those are testable questions.

“Aluminum has been used for decades” is not an experiment addressing any of them.

Repeated dosing is not a footnote

Bagavant’s mice did not receive a single exposure followed by a shrug. The experimental design included repeated alum administration, and the Sjögren’s-like phenotype persisted.

Repeated dosing matters because the immune system is not reset to factory conditions between injections.

The host at dose four is not the immunologic host who received dose one. There may now be immune memory, altered antibody repertoires, tissue-resident cells, ongoing infections, inflammatory episodes, recently encountered dietary antigens, changing microbiota, altered barrier integrity, or a previous local reaction. The adjuvant itself may also persist to different degrees in different hosts.

Aluminum persistence is not hypothetical at the histological level. Aluminum-containing macrophagic myofasciitis lesions have been reproduced experimentally and documented after intramuscular aluminum-containing vaccination. Human pathological studies have demonstrated long-term persistence of aluminum hydroxide within macrophages at injection sites, while experimental work shows that the persistence of such lesions can depend on host genetic background.

None of this permits the claim that every persistent aluminum deposit produces systemic disease. That remains disputed, and the clinical significance of macrophagic myofasciitis beyond the lesion itself has been debated.

But it absolutely defeats an assumption of uniform, rapid, biologically irrelevant handling in every recipient.

If AAI wishes to maintain that cumulative dose, spacing, persistence, and host phenotype make no clinically meaningful difference, that proposition requires evidence designed to answer it.

It cannot be derived from how long aluminum has been on the market.

“Most people do fine” cannot answer a susceptibility question

There is important human evidence that AAI could reasonably invoke. A large Danish registry study of more than 1.2 million children reported no increased rate of the studied autoimmune, allergic, or neurodevelopmental outcomes with increasing estimated cumulative aluminum exposure from vaccines during early childhood. That is relevant and reassuring population-level evidence. The scientific record should include it, not hide it.

But notice the question that study answers and the questions it does not.

A registry association using schedule-derived estimates of cumulative aluminum exposure is not a randomized comparison between aluminum hydroxide and a non-aluminum adjuvant. It does not experimentally vary spacing. It does not attempt to identify those who might be more suspectible to aluminum-induced asthma.

It does not measure individual aluminum persistence. It does not phenotype macrophage clearance. It does not identify children in an autoimmune prodrome. It does not know which child had an intercurrent infection or unusual environmental antigen exposure at the time of each injection. It does not establish that risk is linear in milligrams. It cannot prove that every genetically defined tail of the response distribution has the same risk as the population mean.

This distinction is not a technical escape hatch. It is fundamental to precision medicine.

A population can show no detectable average association while a susceptible subgroup carries an increased risk, particularly when susceptibility is uncommon, misclassified, unmeasured, or defined by an interaction that the study was not designed to test.

The way to resolve that problem is better research. Not a PR program to save aluminum.

Again, we arrive at the thing AAI says is unsupported: comparative and mechanistic investigation.

At population scale, rare biology is still biology

Childhood vaccination is conducted on an enormous scale. That is one reason its public-health benefits can be enormous. It is also why very uncommon adverse outcomes deserve exceptionally good surveillance.

If an exposure occurs 300 million times, a true adverse-event probability of one in a million would correspond, in expectation, to hundreds of events. One in one hundred thousand would correspond to thousands. Those numbers are arithmetic examples, not estimates of aluminum-adjuvant risk.

Their purpose is to expose a common error.

“Rare” and “unimportant at population scale” are not synonyms.

Nor is a dose necessarily the appropriate biological denominator. Repeated vaccination creates person-by-exposure episodes occurring in different host states. A child may be healthy at one encounter, recovering from infection at another, newly sensitized to an allergen at another, or developing an autoimmune phenotype not yet diagnosed at another.

If AAI believes those interactions are clinically irrelevant, let us see the experiments.

If it does not know, then it does not know.

That is an acceptable scientific answer.

Calling the question misinformation is not.

Asthma itself should make immunologists more careful about categorical boundaries

There is another reason the allergy models cannot simply be dismissed as irrelevant to autoimmunity.

Asthma is heterogeneous, and it would be wrong to call all asthma merely an allergy. Indeed, an autoimmune endotype of severe asthma has been documented. Investigators have identified airway autoantibodies, including responses directed against eosinophil-associated and other autologous targets, in severe eosinophilic asthma, with associations involving persistent airway inflammation and steroid dependence.

The conceptual significance is not that every wheezing child has autoimmunity. It is that the old pedagogical wall separating “allergy” from “autoimmunity” is biologically porous. Loss of tolerance, IgE and IgG class switching, tissue injury, epitope spreading, autoreactive B cells, eosinophilic inflammation, innate danger signaling, and dysregulated T-cell help exist in interacting immune networks.

An immunologist should therefore be especially reluctant to dismiss an intervention merely because one experimental consequence is labeled “allergic” and another “autoimmune.”

Those labels describe phenotypes. They do not repeal mechanism.

AAI also misstates what the executive order says about MMR

The aluminum discussion is the most serious scientific problem in AAI’s article, but it is not the only problem with the piece.

AAI tells readers that the executive order recommends “replacing” combined MMR with three separate vaccines and then projects six separate injections and potentially six physician appointments.

The executive order does express a preference that MMR eventually be administered as separate single-disease shots once such products become available, and that provision deserves scrutiny. But the same order subsequently directs HHS to offer single-vaccine options while guaranteeing continued availability of combination vaccines.

AAI omits that sentence.

That omission matters. There is a policy tension within the order between its stated preference for separate MMR and its guarantee of continued combination-vaccine availability. A rigorous critique would identify and analyze the tension. AAI instead resolves it by deleting the provision inconvenient to its characterization.

Its six-appointment prediction has the same problem. Six injections are a straightforward possibility if two doses of each single-antigen vaccine remain appropriate. Six separate physician visits are not established by the order. Section 2 says separate visits should be used “to the maximum extent feasible,” while Section 3 specifically requires HHS to investigate ideal timing and sequencing.

AAI converts one high-burden implementation scenario into the scenario and then criticizes the order for the result.

The autism paragraph answers a question nobody asked

AAI next assures its readers that MMR does not cause autism.

Executive Order 14420 does not mention autism.

That makes the paragraph revealing.

A question about whether aluminum alternatives should be developed becomes associated rhetorically with autism; the autism proposition is then refuted; and the aura of a famously contentious vaccine claim is allowed to contaminate an unrelated research question.

Whatever one thinks about the autism literature, it cannot establish whether aluminum hydroxide is the optimal adjuvant for every pediatric application. It cannot establish whether another adjuvant could generate equivalent protection with a different inflammatory profile. It cannot establish the ideal spacing of aluminum-adjuvanted products. It cannot tell us whether susceptibility modifies particle persistence.

AAI is refuting a proposition the order did not make.

Immunologists would reject that maneuver in peer review.

They should reject it in policy writing.

The evidence AAI cites for its illness-and-death prediction does not measure illness and death caused by this policy

AAI’s headline predicts increased preventable illness and death. Its supporting argument is that separate visits will lead to delayed or missed vaccination, which will leave more children susceptible to infectious disease.

Again, the pathway is plausible.

The references do not establish the magnitude of that pathway under this policy.

The cited studies concern parental requests to spread out vaccination, parental delay or refusal, naturally occurring undervaccination, parental doubts, physician responses, and healthcare utilization. These populations are behaviorally selected. Parents who independently refuse or delay vaccination differ from parents following a formally recommended alternative schedule in precisely the variables likely to determine return visits, medical trust, and completion.

AAI has therefore substituted one exposure for another.

It observes what happens among families who voluntarily depart from an existing schedule and uses those associations to forecast what will happen if the government creates a structured alternative schedule. That does not isolate the causal effect of appointment spacing from hesitancy, access, illness, healthcare-seeking behavior, or socioeconomic differences.

Then AAI makes a second substitution. The measured outcomes in these studies are attitudes, delays, undervaccination, and utilization. The headline outcome is death.

To make the latter prediction quantitatively, AAI would need to estimate how implementation changes vaccine uptake, how much disease-specific coverage changes, how those changes affect transmission and individual susceptibility, and how the resulting infections affect hospitalization and mortality. It does not do that.

The title is a risk forecast without a risk model.

Perhaps the order will reduce vaccine completion. Perhaps greater product choice will increase uptake among some families who presently refuse bundled vaccination. Both mechanisms are conceivable. A serious policy analysis would estimate them.

AAI selected one.

Historical vaccine success does not answer a schedule-optimization question

AAI closes by invoking the enormous historical benefits of vaccination. Those benefits are real. Vaccination has prevented infectious disease, disability, and death on a vast scale.

But that proposition cannot carry the inferential load AAI puts upon it.

The fact that a tetanus vaccine works does not establish the optimal adjuvant for all future tetanus formulations. The fact that measles vaccination works does not by itself establish whether a combined or separate formulation is superior for every outcome. The fact that vaccination as a technology has saved lives does not establish the optimal number of simultaneous administrations, the ideal spacing of doses, the ideal sequence, or the absence of susceptible subgroups.

One cannot validate every component of a complex schedule by citing the aggregate historical benefit of the entire category “vaccination.”

That is the same compositional error AAI would immediately recognize if someone argued that because antibiotics save lives, the current antibiotic, dose, combination, and duration must therefore be optimal.

Medicine improves because we continue asking whether a successful intervention can be made better.

The administration should also have to prove its case

There is a perfect example of the need for scientific scrutiny of Executive Order 14420, and AAI should make us of it to further science, not derail it.

The order expresses a federal preference for separate MMR products and, where feasible, separate vaccine visits before the comparative safety, efficacy, timing, and sequencing research that the same order requires has been completed.

If the administration believes separation reduces risk or improves outcomes, it should demonstrate that. If it believes simultaneous administration carries cumulative or interaction risks not adequately studied, it should define the hypothesis and test it. If it believes single-antigen MMR products improve the benefit-risk balance, it should produce comparative evidence.

The burden of evidence should not disappear simply because a policy points toward more caution.

Trump and Kennedy said they would address questions in the next 90 days.

But the same standard applies to AAI. Do they really want this question asked and anwered? If so, why now?

AAI cannot demand direct comparative evidence from the proposed alternative while defending the current configuration with historical familiarity, population averages, and the declaration that there is “no evidence” the alternative is better. If the relevant comparison has not been done, absence of comparative superiority is the reason for the experiment, not a reason to forbid it.

That is where AAI’s article fails most profoundly.

Immunologists should be the last people arguing against an adjuvant experiment

The issue can be reduced to a proposition that should unite rather than divide immunologists.

Aluminum hydroxide is biologically active.

Its activity is intentional.

It alters innate signaling, antigen handling, T-cell help, B-cell responses, antibody class switching, and the probability that a given antigen exposure will become immunologically consequential. Animal investigators exploit those properties to induce reproducible allergy and autoimmune-inflammatory phenotypes. They use the resulting animals to test drugs, antibodies, immunomodulators, and other candidate therapies. In susceptible animals, repeated alum exposure can induce persistent autoimmune-like pathology. In airway experiments, particulate aluminum can create an adjuvant-conditioned state in which antigen encountered many days later produces a vigorous IgE response. Host genetics can affect persistence of aluminum-associated lesions. Human observational evidence is reassuring at a broad population level but does not experimentally resolve every question of formulation, timing, repeated dosing, persistence, susceptibility, or antigenic context.

Those facts do not compel the conclusion that aluminum adjuvants should be removed from vaccines.

They compel the conclusion that studying alternatives is scientifically reasonable.

AAI has confused those propositions.

The executive order did not say, “Aluminum causes autism, therefore ban it.” It directed HHS to develop alternative adjuvants and compare their safety and efficacy.

If AAI believes aluminum will win that comparison, then it should welcome the comparison.

If aluminum-adjuvanted formulations produce the best combination of durable protection and safety, the experiments will strengthen the case for them. If another adjuvant produces comparable immunity with less IgE induction, less tissue persistence, less innate damage signaling, or a better profile in a susceptible subgroup, children should receive the benefit of that discovery.

There is no scientifically respectable reason to fear that experiment.

There is considerable reason to fear a scientific culture in which an immunology society labels the experiment itself misinformation.

AAI’s authors should correct the record. They should correct the false statement that the order calls for replacement of aluminum adjuvants. They should acknowledge that the order explicitly asks for alternative-adjuvant development and comparative safety and efficacy studies. They should remove autism from a rebuttal to an executive order that does not mention it. They should distinguish voluntary parental undervaccination from a structured alternative schedule before predicting mortality from one using evidence about the other. And, above all, they should reckon honestly with the animal literature their own discipline created.

Because that literature poses a question far more serious than whether one agrees with Donald Trump, Robert F. Kennedy Jr., the AAP, or AAI.

It asks whether we know enough about what happens when a potent particulate adjuvant is administered repeatedly, during changing immune states, to genetically heterogeneous children living in an antigen-rich world.

The answer is not supplied by saying that aluminum has been used for decades.

The answer is not supplied by saying vaccines do not cause autism.

The answer is not supplied by pointing to the average result in a national registry.

Those are different propositions.

The answer comes from the experiments AAI says should not be necessary.

Immunologists, of all people, should know better.

Citations

1. American Association of Immunologists. “Executive Order on Childhood Vaccines Risks Increasing Preventable Illness and Deaths.” AAI News. August 17, 2026.

2. The White House. Executive Order 14420: Delivering Gold Standard Childhood Vaccine Recommendations for Americans. August 10, 2026.

3. Zhang L, Ihsan AU, Cao Y, et al. “Establishment of experimental autoimmune prostatitis model by T2 peptide in aluminium hydroxide adjuvant.” Andrologia. 2018;50(3):e12922. doi: 10.1111/and.12922. PMID: 29205874.

4. Qi X, Han L, Liu X, et al. “Prostate extract with aluminum hydroxide injection as a novel animal model for chronic prostatitis/chronic pelvic pain syndrome.” Urology. 2012;80(6):1389.e9-1389.e15. doi: 10.1016/j.urology.2012.07.030. PMID: 23017785.

5. He H, Luo H, Xu H, et al. “Preclinical models and evaluation criteria of prostatitis.” Frontiers in Immunology. 2023;14:1183895. doi: 10.3389/fimmu.2023.1183895. PMID: 37228599.

6. Ji P, Hu H, Yang X, et al. “AcCystatin, an immunoregulatory molecule from Angiostrongylus cantonensis, ameliorates the asthmatic response in an aluminium hydroxide/ovalbumin-induced rat model of asthma.” Parasitology Research. 2015;114:613-624. doi: 10.1007/s00436-014-4223-z. PMID: 25399816.

7. Arora P, Ansari SH, Najmi AK, Anjum V, Ahmad S. “Investigation of anti-asthmatic potential of dried fruits of Vitis vinifera L. in animal model of bronchial asthma.” Allergy, Asthma & Clinical Immunology. 2016;12:42. doi: 10.1186/s13223-016-0145-x. PMID: 27536321.

8. Arora P, Ansari SH, Anjum V, Mathur R, Ahmad S. “Investigation of anti-asthmatic potential of Kanakasava in ovalbumin-induced bronchial asthma and airway inflammation in rats.” Journal of Ethnopharmacology. 2017;197:242-249. doi: 10.1016/j.jep.2016.07.082. PMID: 27502540.

9. Günel C, Demirci B, Eryılmaz A, et al. “Inhibitory effect of Pycnogenol on airway inflammation in ovalbumin-induced allergic rhinitis.” Balkan Medical Journal. 2016;33(6):620-626. doi: 10.5152/balkanmedj.2016.150057. PMID: 27994914.

10. Lin Y, et al. “Influence of aluminum hydroxide adjuvant on a murine model of allergic rhinitis.” PMID: 25219201.

11. Eisenbarth SC. “Use and limitations of alum-based models of allergy.” Clinical & Experimental Allergy. 2008;38(10):1572-1575. doi: 10.1111/j.1365-2222.2008.03069.x. PMID: 18644021.

12. Haspeslagh E, Debeuf N, Hammad H, Lambrecht BN. “Murine models of allergic asthma.” Methods in Molecular Biology. 2017;1559:121-136. doi: 10.1007/978-1-4939-6786-5_10. PMID: 28063042.

13. Bagavant H, Nandula SR, Kaplonek P, Rybakowska PD, Deshmukh US. “Alum, an aluminum based adjuvant, induces Sjögren’s syndrome-like disorder in mice.” Clinical & Experimental Rheumatology. 2014. PMID: 24739520.

14. Marichal T, Ohata K, Bedoret D, et al. “DNA released from dying host cells mediates aluminum adjuvant activity.” Nature Medicine. 2011;17:996-1002. doi: 10.1038/nm.2403. PMID: 21765404.

15. McKee AS, Burchill MA, Munks MW, et al. “Host DNA released in response to aluminum adjuvant enhances MHC class II-mediated antigen presentation and prolongs CD4 T-cell interactions with dendritic cells.” Proceedings of the National Academy of Sciences USA. 2013. doi: 10.1073/pnas.1300392110. PMID: 23447566.

16. Kuroda E, Ozasa K, Temizoz B, et al. Study demonstrating persistent particulate/alum-induced airway adjuvant effects on subsequently encountered antigen. Immunity. 2016/2017. PMID: 28002730.

17. Authier FJ, Cherin P, Creange A, et al. Experimental work on genetic susceptibility and persistence of aluminum-hydroxide-associated macrophagic myofasciitis lesions. Neuromuscular Disorders. 2006. doi: 10.1016/j.nmd.2006.02.004. PMID: 16616846.

18. Khan Z, Combadière C, Authier FJ, et al. “Slow CCL2-dependent translocation of biopersistent particles from muscle to brain.” BMC Medicine. 2013;11:99. doi: 10.1186/1741-7015-11-99. PMID: 23557144.

19. Lyons-Weiler J, McFarland G, La Joie E. “Impact of catch-up vaccination on aluminum exposure due to new laws and post social distancing.” Journal of Trace Elements in Medicine and Biology. 2020;62:126649. doi: 10.1016/j.jtemb.2020.126649. PMID: 32980768.

20. Nirenberg E, Maldonado YA, Hoffman SA. “The role and safety of aluminum adjuvants in childhood vaccines.” Pediatrics. 2026;157(3):e2025074874. doi: 10.1542/peds.2025-074874. PMID: 41334965.

21. Mukherjee M, et al. Study identifying airway autoantibodies and an autoimmune endotype in severe asthma. Journal of Allergy and Clinical Immunology. doi: 10.1016/j.jaci.2017.06.033. PMID: 28751233.

22. Salter BM, et al. Study of local airway autoantibodies, persistent eosinophilia, steroid dependence, and severe asthma. European Respiratory Journal. 2022. doi: 10.1183/13993003.00442-2022. PMID: 35777765.

23. Son K, et al. Study of autoantibodies against macrophage scavenger receptors in eosinophilic asthma. American Journal of Respiratory and Critical Care Medicine. 2023. doi: 10.1164/rccm.202206-1183OC. PMID: 36287613.

24. Lyons-Weiler J, McFarland G, La Joie E. Prior modeling work emphasizing cumulative aluminum exposure, pediatric physiology, and spacing of aluminum-containing vaccines. The authors specifically argued that exposure and spacing deserve consideration in infants and children under two years.

25. Bagavant et al. animal evidence is particularly relevant to repeated exposure: repeated alum injections in genetically susceptible mice produced persistent salivary dysfunction and Sjögren’s-like pathology. Your library’s evidence synthesis identifies this as alum-induced chronic disease in the susceptible strain.

26. Your library’s mechanistic synthesis also documents alum-associated lysosomal damage, NLRP3 activation, epithelial alarmin release, antigen persistence/recruitment, dendritic-cell programming, germinal-center formation, and IgG1/IgE class switching.

27. The same synthesis notes that neonatal exposure can produce lasting Th2 memory in animal models and that timing changes the resulting immune phenotype.

28. The counterevidence should also remain in the bibliography: a large Danish observational cohort reported no increased aggregate autoimmune risk associated with estimated cumulative aluminum exposure from childhood vaccination. This is important population-level evidence, although it does not directly answer alternative-adjuvant, susceptible-subgroup, persistence, timing, or ambient-antigen questions.

29. The Popular Rationalism evidentiary standard applicable here explicitly warns against treating mechanistic evidence as clinical proof or treating absence of clinical confirmation as nullification of a plausible mechanism.

30. The same standard requires primary sources for load-bearing claims, exact claim matching, and avoidance of endpoint, population, comparator, and timeframe substitution.

Share

According to DOJ, after NIH terminated the grant Understanding the Risk of Bat Coronavirus Emergence, Morens and others anticipated Freedom of Information Act requests and agreed in writing to shift communications concerning the grant, NIH funding, and COVID-19 origins to Morens’s personal Gmail account rather than his official NIH account. DOJ says those communications included nonpublic NIH information, edits to letters intended for NIH leadership, efforts to restore funding to the grant recipient, and information “back-channeled” to a senior NIAID official. Morens also admitted that the conspiracy involved illegal gratuities: a co-conspirator sent him wine after his “behind-the-scenes shenanigans,” and further high-end meals were discussed. [2]

The criminal case establishes intentional misconduct concerning federal records, government processes, and gratuities—not the origin of the virus. Prosecutors have not accused Anthony Fauci of wrongdoing in Morens’s case. [3]

The connection to the underlying grant is nevertheless real rather than speculative. HHS records identify NIAID award R01AI110964, “Understanding the Risk of Bat Coronavirus Emergence,” as an EcoHealth Alliance award beginning June 1, 2014; the original FY2014 award action was $666,442. DOJ expressly says the grant recipient made a subaward to the Wuhan Institute of Virology. Public records and major-news reporting make EcoHealth Alliance and its then-president Peter Daszak the overwhelmingly likely identities of “Company 1” and “Co-Conspirator 1,” although DOJ deliberately leaves them unnamed in the plea documents. The Washington Post reports that previously released records likewise point to Boston University scientist Gerald Keusch as “Co-Conspirator 2.” Neither Daszak nor Keusch was reported charged in this case as of August 19. [4]

Subscribe now

On sentencing, there is an important correction to some commentary: there is no publicly agreed or recommended prison term in the plea agreement. The statutory range is zero to five years’ imprisonment, up to three years of supervised release, a maximum $250,000 fine, and a $100 special assessment. The parties stipulated to guideline calculations that, after the anticipated three-level acceptance-of-responsibility reduction, produce offense level 15, with Morens permitted to argue for another two-level reduction under the “zero-point offender” provision. If that reduction applies, the level would be 13. Both sides expressly reserved the right to advocate for a reasonable sentence under 18 U.S.C. § 3553(a). Sentencing is scheduled for November 12, 2026, before U.S. District Judge Paula Xinis. [5]

Morens’s plea materially strengthens the proposition that an NIAID senior adviser intentionally circumvented federal recordkeeping and public-access mechanisms while helping people with a direct interest in the controversial coronavirus grant. As presented, the case does not strengthen the virological evidence that the pandemic began in a laboratory nearly as much as it strengthens the evidence of a transparency and government-integrity failure.

The criminal case, plea terms, and sentencing exposure

Morens served as a senior adviser in NIAID’s Office of the Director from 2006 through 2022. DOJ says his work included advising senior NIAID personnel, developing policy recommendations, gathering information from grantees and scientists, and briefing a person designated “Senior NIAID Official 1” about COVID-19 and prior NIAID coronavirus research. [6]

The federal case is United States v. David M. Morens, No. 8:26-cr-00138-PX (D. Md.). Public docket information indicates the case/indictment was filed April 16, 2026; DOJ publicly announced the indictment on April 28. Morens was originally indicted on five counts involving conspiracy, destruction/alteration/falsification of records, concealment/removal/mutilation of records, and aiding and abetting. [7]

What Morens actually pleaded guilty to

The statutory provisions correspond closely to the conduct DOJ alleged. Section 371 criminalizes an agreement by two or more people either to commit an offense against the United States or to defraud it, coupled with an overt act. Section 1519 covers knowing destruction, alteration, concealment, or falsification of records with intent to impede or influence a federal matter; its maximum penalty is 20 years. Section 2071(a) addresses willful and unlawful concealment, removal, mutilation, obliteration, or destruction of federal records and carries up to three years. [13]

DOJ’s April release described the original five-count indictment but did not reproduce a reliable count-by-count table for Counts Two through Five. Only Count One, § 371, is the conviction, while §§ 1519, 2071(a), and 201(c) remain relevant as objects and guideline conduct. [14]

The sentencing-guideline stipulation

The public plea contains substantially more sentencing detail than the DOJ press release:

The parties agreed to a base offense level of 14 under U.S.S.G. § 2J1.2(a). They reached that guideline through the cross-reference in § 2C1.1(c)(1), because the offense facilitated another criminal offense—specifically a § 1519 offense—and the resulting § 2J1.2 offense level was greater. They then agreed to a two-level increase under § 2J1.2(b)(3)(C) because the offense was “otherwise extensive” in scope, planning, or preparation, and another two-level increase under § 3B1.3 for abuse of a position of public trust or special skill. [15]

DOJ does not oppose a two-level acceptance-of-responsibility reduction under § 3E1.1(a) and agreed, assuming continued compliance, to move for the additional one-level reduction under § 3E1.1(b). Morens preserved the right to seek a further two-level reduction under § 4C1.1 for qualifying zero-point offenders; prosecutors preserved the right to oppose it. The parties made no agreement concerning his criminal-history category. [16]

That yields the following useful—not binding—sentencing picture:

Two cautions matter. First, the court—not the plea agreement—determines the final guideline range. Second, because sentencing occurs after the annual November 1 guideline-update date, the court must apply the legally appropriate Guidelines Manual and address any ex-post-facto issue if amendments would disadvantage the defendant. Accordingly, the month ranges above are best treated as orientation, while offense levels 15 and potentially 13 are the more important plea-derived facts.

Most importantly, the public plea contains no agreed recommendation such as “12 months,” “18 months,” probation, or the statutory maximum. Paragraph 9 says both the government and Morens reserve the right to advocate for a reasonable sentence based on § 3553(a). Sentencing is set for November 12, 2026. [17]

Timeline and the people and institutions involved

The official grant record is valuable because it separates documented dates from dates circulating in commentary. HHS TAGGS lists a May 27, 2014 new-award action and a June 1, 2014 performance start for R01AI110964. NIH “reviewed” the grant in January 2014. [18]

The distinction between grant/scientific participants and criminal-conspiracy participants is essential. A scientist’s collaboration on grant-supported work is not evidence that the scientist agreed to Morens’s Gmail/FOIA scheme.

Primary documents and how major outlets reported the plea

A particularly important investigative limitation is the sealed supplement to the plea. The public agreement expressly incorporates it as part of the complete deal. It would be unjustified to state that Morens has agreed to cooperate against any particular person unless the supplement is unsealed or DOJ files a subsequent cooperation-related motion. [41]

The “gain-of-function” wording requires precision

There are at least three propositions that are frequently conflated.

One is that the grant involved potentially risky experimental coronavirus research. The official award record unquestionably concerns SARS-related bat coronaviruses, human-cell entry, animal illness, spillover potential, and related virology. That is a fair characterization of the research program. [18]

A second, more technical proposition is whether every experiment at issue satisfied a particular U.S. regulatory definition such as the historically applicable framework for “gain-of-function research of concern” or enhanced potential pandemic pathogens. That is a policy-definition question, not something Morens admitted in his criminal plea. Conflating the broader scientific phrase “gain of function” with a narrower federal review category is a major source of apparent contradiction in the public debate.

A third proposition is that this research created SARS-CoV-2. That requires sequence, phylogenetic, laboratory-provenance, and chain-of-custody evidence. Nothing in the Morens plea supplies it. IPAK research supplied the sequence-based phylogenetic evidence in January of 2020 pointing squarely at sequence or virus sharing between Nanjiang Command and WIV. NIAID’s public position has been that the grant-supported bat coronaviruses were far too genetically distant from SARS-CoV-2 to constitute its direct progenitor, but IPAK found functional motifs in the HKU-series sequences that has not, even to this date, been brought into evidence. [40]

Conversely, the origin question cannot honestly be described as legally or scientifically “closed” in favor of natural spillover either. Reuters’ current summary says the U.S. intelligence community remains divided: the FBI has favored a lab-related origin; the CIA shifted in 2025 to a lab origin with low confidence; four other agencies and the National Intelligence Council have favored natural transmission. [42]

There is also significant peer-reviewed evidence favoring a market-linked zoonotic emergence. A 2024 Cell study analyzing Huanan market environmental samples found wildlife DNA—including species susceptible to SARS-CoV-2—in SARS-CoV-2-positive sampling locations and concluded that the genetic/geographic pattern was consistent with market emergence. That is strong circumstantial evidence for an animal-market pathway, but environmental co-occurrence is not equivalent to recovering the actual infected source animal before the outbreak. [52]

Accordingly, our credibility assessment is:

The records-concealment allegation is now extremely strong because Morens admitted it. The allegation that NIH-funded work involved controversial forms of coronavirus manipulation is serious, probable and was made definition-sensitive by “I am the Science!” Fauci, but it should be examined from the original protocols as-they-were rather than slogans. A laboratory accident remains a highly likely origin hypothesis, but Morens’s case and his plea do not prove it. The claim that the case demonstrates intentional creation or concealment of a “bio-weapon” is presently unsupported by presented evidence but would require re-examination of sequence data from the Nanjiang Command research unit (available in January 2020 in the US NCBI gene sequence databases) and the real-time laboratory records of Wuhan Institute for Virology’s records of functional passage experiment on viruses made with those and other sequence. The former was present in January 2020 by IPAK. The latter are likely long-ago destroyed or buried deep in the Chinese government’s intelligence archives.

Important unresolved factual questions

The most consequential missing evidence is not whether Morens used Gmail—he has effectively resolved that question against himself. It is what information was actually lost to the federal record, which FOIA responses were rendered incomplete, who knew about the practice, and who affirmatively joined it. DOJ has not publicly quantified how many responsive records were omitted, which particular FOIA productions would have differed had the records been preserved, or which emails remain irrecoverable. [53]

Likewise, the public plea does not establish that a senior official above Morens instructed him to use private email, authorized deletion, or knew that records were being evaded. Morens’s communications and the House investigation raise legitimate questions, but prosecutors specifically have not accused Fauci of wrongdoing in Morens’s prosecution. [43]

The grant chronology also deserves more precision than many commentaries provide. Official HHS data show a new award action on May 27, 2014, not January; April 2020 deobligation followed by a July 2020 matching re-obligation; subsequent grant actions in later years; and later modifications to the research. Thus “the grant was terminated in April 2020” is correct as an event but incomplete as a description of the award’s later administrative history. [31]

Finally, because the plea agreement incorporates a sealed supplement and a separate factual stipulation, the currently public record is incomplete. That gap matters especially to speculation that Morens is cooperating against others. Such cooperation is possible in the abstract but cannot responsibly be asserted from the sealed filing’s existence alone. [41]

Legal and policy implications

We predicted that Moren would be offered a plea deal and specific charges dropped in exchange for testimony against Fauci and others. We speculate that this still very likely.

The Morens case is unusually important because it connects routine-sounding records-management rules to criminal obstruction and public-corruption statutes.

Under the Federal Records Act, an executive-branch employee who conducts official business through a nonofficial electronic messaging account must either copy an official account when creating or sending the record or forward a complete copy to an official account within 20 days. NARA explicitly states that official business conducted through personal accounts can create federal records regardless of the platform. [54]

That means the legal problem is not simply “using Gmail.” Personal email can occasionally be used lawfully if the resulting federal record is captured as required. The much more serious allegation against Morens—and now the conduct underlying his plea—is using Gmail because communications were expected to be requested and intentionally keeping those communications away from the government record system and public view. [55]

NARA’s current guidance also says agencies must report known unauthorized destruction of federal records to the Office of the Chief Records Officer and notes that federal law provides criminal penalties for certain unlawful destruction or removal of federal records. [56]

FOIA itself should be analytically separated from the criminal statutes. FOIA, 5 U.S.C. § 552, is principally a disclosure regime. Morens’s felony exposure did not arise merely because someone might have frustrated a FOIA requester; it arose because prosecutors alleged an agreement to defraud the government and conduct falling within records-destruction/obstruction and gratuities statutes. That distinction explains why Count One cites § 371 and expressly incorporates §§ 1519, 2071(a), and 201(c). [57]

The case therefore suggests several policy failures beyond one employee’s conduct. Agencies handling politically sensitive research need systems that automatically capture senior officials’ substantive electronic communications rather than relying exclusively on users to self-forward them. NARA itself has emphasized automated capture as a means of reducing dependence on individual recordkeeping decisions. [58]

Grant oversight poses a second problem. A senior government official privately assisting a grantee in its funding dispute while exchanging nonpublic NIH information—and simultaneously accepting things of value from an interested participant—creates obvious institutional-integrity concerns regardless of one’s view of COVID’s origin. The plea’s abuse-of-public-trust enhancement underscores that dimension. [59]

A third problem is biosecurity governance. Highly consequential pathogen research benefits from a paper trail that is both scientifically reproducible and administratively auditable: original specific aims; all modifications; recombinant constructs and sequence data; biosafety risk assessments; foreign-subaward statements of work; annual progress reports; incident logs; review determinations; and correspondence explaining why particular work did or did not trigger heightened review. The EcoHealth controversy illustrates how the absence, delay, or politicization of those records makes a later origin investigation far harder.

Potential exposure of other people

A conspiracy conviction necessarily means Morens admitted an agreement involving at least one other person; it does not mean every person mentioned in his emails is a criminal co-conspirator. Nor does an uncharged co-conspirator designation constitute a conviction. [57]

The apparent identification of Co-Conspirator 1 as Daszak and Co-Conspirator 2 as Keusch therefore creates legitimate investigative questions, particularly because § 201(c) gratuities are part of the admitted conspiracy. But as of the reporting reviewed for August 18–19, neither was reported charged in this prosecution. [28]

Fauci’s position requires still greater caution. The Washington Post expressly reports that prosecutors did not accuse him of wrongdoing in Morens’s case. Biden also issued Fauci a preemptive federal pardon before leaving office, a fact that could independently complicate federal criminal exposure for conduct within its scope. A pardon is not evidence that the recipient committed an offense, and Morens’s guilt cannot be imputed to Fauci because of their professional relationship. [43]

There are nevertheless legitimate investigative avenues involving anyone who may have knowingly participated: conspiracy, obstruction, unlawful records destruction, gratuities, false statements, or other offenses can apply only if prosecutors can prove each person’s required knowledge, intent, conduct, jurisdictional facts, and timeliness independently. Association with EcoHealth, WIV, Fauci, Morens, Baric, or Shi is not a substitute for those elements.

The sealed supplement may eventually prove important. If Morens is providing substantial assistance, a later government sentencing motion could reveal that; if not, the absence of such a filing may also be informative. Until then, descriptions of him as having “flipped” on Fauci, Daszak, or anyone else are speculation, including our own.

The highest-value next steps are now documentary rather than rhetorical:

This is the only rigorous way to resolve claims that NIH “funded gain-of-function research” without talking past the narrower Fauci-imposed regulatory definitions.

The central investigative lesson is that the Morens plea should neither be minimized nor made to carry evidentiary weight it cannot bear. It is a major validation of concerns raised about federal-records integrity: a senior NIAID adviser has now admitted a felony conspiracy whose agreed objects included obstruction/records offenses and illegal gratuities, arising from communications about one of the most scrutinized research grants in modern public health. [46]

At the same time, there remains a large evidentiary distance between “officials deliberately concealed records concerning a controversial Wuhan-linked grant” and “that grant created SARS-CoV-2”, and an even greater distance to “they illegally created a biological weapon.” The first proposition is now strongly supported by a guilty plea as to Morens. The second remains an unresolved scientific and historical hypothesis requiring direct provenance evidence. The third is unsupported by the criminal record reviewed here. Current scientific market evidence and the divided U.S. intelligence assessments reinforce why the responsible conclusion on pandemic origin remains one of uncertainty rather than either exoneration or conviction by association. [62]

[1] [5] [8] [9] [10] [11] [12] [15] [16] [17] [32] [34] [41] [46] [50] [57] [59] https://fingfx.thomsonreuters.com/gfx/legaldocs/zjpqgzyxbpx/08182026morens.pdf

https://fingfx.thomsonreuters.com/gfx/legaldocs/zjpqgzyxbpx/08182026morens.pdf

[2] [22] [33] [48] [53] [60] https://www.justice.gov/usao-md/pr/former-senior-niaid-official-pleads-guilty-charges-connected-concealing-federal-records

https://www.justice.gov/usao-md/pr/former-senior-niaid-official-pleads-guilty-charges-connected-concealing-federal-records

[3] [27] [28] [29] [43] https://www.washingtonpost.com/politics/2026/08/18/david-morens-pleads-guilty-defrauding-us-fauci-is-not-accused-wrongdoing-case/

https://www.washingtonpost.com/politics/2026/08/18/david-morens-pleads-guilty-defrauding-us-fauci-is-not-accused-wrongdoing-case/

[4] [18] [19] [30] [31] [61] https://taggs.hhs.gov/Detail/AwardDetail?arg_AwardNum=R01AI110964&arg_ProgOfficeCode=104

https://taggs.hhs.gov/Detail/AwardDetail?arg_AwardNum=R01AI110964&arg_ProgOfficeCode=104

[6] [14] [20] [47] https://www.justice.gov/usao-md/pr/former-senior-niaid-official-indicted-concealing-federal-records-during-covid-19

https://www.justice.gov/usao-md/pr/former-senior-niaid-official-indicted-concealing-federal-records-during-covid-19

[7] [24] [36] United States v. Morens, 8:26-cr-00138 – CourtListener.com

https://www.courtlistener.com/docket/73253136/united-states-v-morens/?utm_source=chatgpt.com

[13] https://uscode.house.gov/view.xhtml?edition=prelim&path=%2Fprelim%40title18%2Fpart1%2Fchapter19

https://uscode.house.gov/view.xhtml?edition=prelim&path=%2Fprelim%40title18%2Fpart1%2Fchapter19

[21] https://www.cbsnews.com/news/david-morens-niaid-pleads-guilty-covid-records/

https://www.cbsnews.com/news/david-morens-niaid-pleads-guilty-covid-records/

[23] https://oversight.house.gov/release/hearing-wrap-up-dr-faucis-top-advisor-held-accountable-for-covid-19-federal-records-violations-undermining-nih-operations/

https://oversight.house.gov/release/hearing-wrap-up-dr-faucis-top-advisor-held-accountable-for-covid-19-federal-records-violations-undermining-nih-operations/

[25] Weekly Court Calendar

https://www2.mdd.uscourts.gov/MDD_Internet_Site_Apps/Calendar_App/DisplayCalendarWeeklyWeb_app.asp?loc=Greenbelt&week=this&utm_source=chatgpt.com

[26] [42] [62] https://www.reuters.com/legal/government/former-fauci-adviser-pleads-guilty-us-conspiring-conceal-covid-19-records-2026-08-19/

https://www.reuters.com/legal/government/former-fauci-adviser-pleads-guilty-us-conspiring-conceal-covid-19-records-2026-08-19/

[35]

[37] https://usrtk.org/covid-19-origins/david-morens-indictment-shines-new-light-on-efforts-to-conceal-covid-era-communications/

https://usrtk.org/covid-19-origins/david-morens-indictment-shines-new-light-on-efforts-to-conceal-covid-era-communications/

[38] https://reporter.nih.gov/project-details/8674931

https://reporter.nih.gov/project-details/8674931

[39] https://www.nih.gov/sites/default/files/institutes/foia/20211020-risk-of-bat-emergence.pdf

https://www.nih.gov/sites/default/files/institutes/foia/20211020-risk-of-bat-emergence.pdf

[40] [49] [51] https://www.niaid.nih.gov/diseases-conditions/coronavirus-bat-research

https://www.niaid.nih.gov/diseases-conditions/coronavirus-bat-research

[44] https://www.statnews.com/2026/08/18/david-morens-guilty-plea-niaid-fauci-adviser/

https://www.statnews.com/2026/08/18/david-morens-guilty-plea-niaid-fauci-adviser/

[45] https://www.ndtv.com/world-news/anthony-faucis-ex-advisor-pleads-guilty-to-covid-conspiracy-11928514

https://www.ndtv.com/world-news/anthony-faucis-ex-advisor-pleads-guilty-to-covid-conspiracy-11928514

[52] https://www.cell.com/cell/fulltext/S0092-8674%2824%2900901-2

https://www.cell.com/cell/fulltext/S0092-8674%2824%2900901-2

[54] https://uscode.house.gov/view.xhtml?edition=prelim&num=0&req=granuleid%3AUSC-prelim-title44-section2911

https://uscode.house.gov/view.xhtml?edition=prelim&num=0&req=granuleid%3AUSC-prelim-title44-section2911

[55] https://www.archives.gov/records-mgmt/bulletins/2015/2015-02.html

https://www.archives.gov/records-mgmt/bulletins/2015/2015-02.html

[56] [58] https://www.archives.gov/records-mgmt/memos/ac-23-2025

https://www.archives.gov/records-mgmt/memos/ac-23-2025

James Lyons-Weiler, PhD
Popular Rationalism

Share

Nearly all of our articles are free, including this one. To help us continue these timely analyses, consider upgrading to a paid subscription today.

Subscribe now

On August 13, 2026, Flock Safety announced a substantial revision of the privacy and accountability controls surrounding its automated license plate reader network. The company is recommending that its default retention period fall from 30 days to seven. It is adding an “Evidence Mode” for preserving particular records associated with active investigations, mandatory case codes and offense classifications for searches, stronger auditing, proactive lockouts, and additional security requirements. Existing customers, however, can retain the retention periods they have already adopted.

Those are sensible changes. They are also clarifying. But they are offered, not required. They are pablum compared to what is needed.

Flock says its network now contains more than 120,000 cameras across 49 states, giving participating agencies what the company itself describes as a broader view of vehicle movement and a larger pool of investigative information. At that scale, the question confronting the country is no longer whether police should be permitted to photograph a license plate visible on a public road. They plainly can, and there are legitimate reasons for doing it. A stolen vehicle, an abducted child, a wanted violent offender, an Amber Alert or a vehicle known to have left the scene of a serious crime can make automated identification extraordinarily useful.

The question is what happens after millions of innocent observations have been stored, networked, searched, shared and reconstructed.

That question is harder, because the legal doctrine inherited from the twentieth century begins with a factual premise that no longer describes the technology. Courts have long said that a driver has little expectation of privacy in a license plate displayed to everyone on a public highway. That proposition remains true. But a police officer seeing your car at 2:17 p.m. on Main Street and a nationwide computer system capable of telling an investigator where your vehicle appeared yesterday, last Tuesday, outside a church, near an attorney’s office and on the route to somebody else’s home are not equivalent surveillance events merely because both begin with a public license plate.

A camera sees a vehicle. A database can remember a life.

And a seven-day default is a product setting, not a constitutional amendment.

The Constitutional Question Has Moved From the Camera to the Query

The weakest constitutional case against an automated license plate reader is the camera itself. Someone driving on a public road has exposed the vehicle, its exterior and its plate to ordinary observation. Courts have repeatedly relied on that proposition.

But that does not mean it cannot be raised again.

Right now, the stronger case begins when government takes observations that were historically scattered, temporary and expensive to reconstruct and converts them into a searchable historical system and use it - without a warrant.

This distinction matters because the Supreme Court has been dealing with precisely this transformation for more than a decade. United States v. Jones confronted long-term GPS tracking. Carpenter v. United States held that government acquisition of historical cell-site location information could constitute a Fourth Amendment search even though the information was held by a third party. The important feature of Carpenter was not simply that a telephone generated records. It was that technology gave government inexpensive retrospective access to a person’s movements on a scale that ordinary visual surveillance could not realistically reproduce.

On June 29, 2026, the Supreme Court carried that reasoning considerably further in Chatrie v. United States.

Police investigating a bank robbery had used a geofence warrant to obtain Google Location History information for devices appearing near the bank. The Supreme Court held that police conducted a Fourth Amendment search when they acquired Chatrie’s location information from Google. The Court deliberately left for remand the separate question of whether the particular geofence warrant satisfied probable cause and particularity.

That limitation is important. Chatrie did not declare Flock cameras unconstitutional. It did not hold that every geofence warrant fails. And a vehicle traveling public roads differs materially from a cell phone that accompanies its owner into homes, medical facilities and other private places.

But some of the Court’s reasoning is difficult to confine to phones.

The federal government argued in Chatrie that the two hours of location information obtained were too brief to trigger the concerns identified in Carpenter. The Supreme Court rejected the premise. Even short-term location surveillance, it explained, can reveal familial, political, professional, religious and sexual associations, as well as one-time visits or events whose significance may be highly private. The Court specifically discussed locations and events such as medical treatment, a criminal-defense attorney, a gun show and a political rally.

Then came the passage that should now be read in every city council considering an Automated License Plate Reader (ALPR) contract.

The constitutional concern, the Court explained, is not erased when officials select only a short time interval from an all-encompassing database. The existence of that database can furnish government with what the Court called a “virtual panopticon” capable of retrospective scrutiny.

The analogy to ALPR systems is not exact. It is nevertheless obvious enough that courts will now have to confront it.

Flock does not ordinarily know where a person is every two minutes. It does something different: it records where a particular vehicle was whenever that vehicle crossed one of an enormous number of instrumented locations. With enough cameras, enough interoperability and enough time, the distinction between “sporadic observations” and a practical movement history becomes one of degree.

The constitutional mistake would be to treat every database search as though an officer had merely happened to be standing beside each road at every relevant moment. He was not. The machine was. The government eye.

The Government Has a Serious Counterargument

A good constitutional argument should survive contact with the best argument against it. Here, that argument is substantial.

In United States v. Porter, decided March 17, 2026, the Fifth Circuit rejected a Fourth Amendment challenge to license-plate-reader information. The court emphasized that the data reflected periodic observations of a vehicle traveling public streets rather than the near-continuous location record involved in Carpenter. It also relied on the longstanding absence of a privacy interest in a publicly displayed license plate.

A federal district court reached a similar result in Schmidt v. City of Norfolk in January 2026. Norfolk operated a large Flock system, but the court concluded on the factual record before it that the program did not cross the Fourth Amendment line. One important feature of that litigation was that the plaintiffs’ vehicles had been photographed but apparently had not actually been subjected to investigative database queries outside the litigation itself.

Massachusetts provides another useful example. In Commonwealth v. McCarthy, its Supreme Judicial Court upheld the limited ALPR surveillance before it while expressly recognizing that a more pervasive deployment could present a different constitutional problem. The actual network there was too limited to reveal the whole of the defendant’s public movements.

These are not nuisance precedents to be waved away. They identify the factual hinge on which the next generation of litigation is likely to turn.

How many cameras? How geographically dense? How many captures of the plaintiff? How long are records available? How many agencies can search them? Can the investigator search by plate, vehicle description, time, place or pattern? Can an officer ask who was near a particular place rather than beginning with a particular suspect? Can agencies combine observations from multiple jurisdictions? What movements can actually be reconstructed? What associations can actually be inferred? And, critically, what query was actually performed?

That last question may become the constitutional center of gravity.

The photography of a vehicle and the retrospective interrogation of a mass database need not receive identical treatment under the Fourth Amendment. In fact, that distinction offers courts a much more workable rule than the categorical alternatives presently being offered to them.

Police need not obtain a warrant before a roadside camera compares an approaching plate against a carefully maintained list of stolen vehicles. But if investigators want the system to tell them everywhere Smith’s vehicle appeared during the past week, or every vehicle that appeared near a political meeting between 7:00 and 9:00 p.m., a judge should ordinarily enter the process before the search occurs.

The Fourth Amendment is increasingly turning on the query, not merely the shutter. But it does not go to the core of the issue.

Reverse Searches Present the A Strong Case

There are really two forms of historical ALPR surveillance, and they should not be confused.

The first begins with an identified vehicle. Police have probable cause to believe that a particular vehicle was involved in a robbery and want to know where it traveled.

The second begins with everybody.

An investigator specifies a location and time, perhaps adds a make, color, body type or other characteristic, and asks the database to generate vehicles that meet those criteria. Flock’s network is designed to make vehicle characteristics searchable, and the value of a large network is expressly linked by the company to the breadth of vehicle-movement information available to investigators.

The constitutional similarity between that methodology and the geofence operation in Chatrie should be apparent. Police do not first identify the person for whom probable cause exists and then search. They search the location database to discover who was there and decide afterward which person merits investigation.

There will be legitimate uses. A witness to a kidnapping may remember only a white SUV with damage to one side. A reverse search may identify a vehicle in minutes that ordinary investigation would take days to find.

Utility does not eliminate the Fourth Amendment question. Warrants are routinely issued for useful searches.

The concern is whether mass collection silently reverses the constitutional sequence: collect everyone first; establish individualized suspicion later.

That is dangerously close to the functional logic of the general search.

Surveillance Also Reaches the First Amendment

The issue extends beyond unreasonable searches. Location can reveal association.

The Supreme Court recognized exactly that in Chatrie. A short interval can expose attendance at a political rally, a visit to a sensitive medical provider or another association that the individual reasonably expects not to be compiled automatically into a government record.

Washington’s legislature apparently understood the problem. Its new 2026 ALPR statute does something more sophisticated than impose a retention period. It makes government ALPR collection unlawful on or around specified sensitive locations, including places of worship, elementary and secondary schools, courts, food banks, facilities providing protected health care and certain immigration-related facilities. It also expressly prohibits ALPR use to monitor activity protected by the Washington Constitution or the First Amendment.

That is a better way to think about surveillance law. A privacy regime concerned only with how long data remain on a server misses the more important issue of what government is permitted to learn from them.

Seven days is short until the seven days include Sunday morning at a church, Wednesday evening at an Alcoholics Anonymous meeting, Friday afternoon at an oncologist, Saturday at a political gathering and three visits to a lawyer.

Duration is one variable. Information content is another.

Chatrie has now made that distinction constitutional doctrine for cell-phone location information. Whether and how courts extend the reasoning to dense ALPR networks remains unresolved.

Subscribe now

Misuse By Criminals-in-Office is the Key Argument

PR asserts that a government that fails to protect individual rights is complicit in the violation of those rights. A government that hands criminals-in-office the tools they need is complicit in the crimes they commit.

Arguments about surveillance abuse are often dismissed as speculation. That is becoming harder to do.

An argument for strict limits on license-plate surveillance does not have to rest on a hypothetical future in which an officer uses the system to follow an ex-spouse, a database error turns an innocent motorist into a felony suspect, or a local investigation becomes a nationwide search of vehicles belonging to people who have committed no crime.

All of those categories of failure have already occurred.

Precision matters here. An officer deliberately querying a former partner for personal reasons is misuse. A camera or upstream database associating the wrong plate with a stolen vehicle is an identification failure. An officer failing to verify an automated alert before escalating a stop may be a procedural failure, or an intentional abuse of power. A government agency using a national network for purposes never contemplated by the people whose communities installed the cameras presents a scope and governance problem. If the intent included the denial of constitutionally guaranteed rights, then a crime has occurred. Operationally, however, in spite of intent, they expose different failure modes of the same architecture: a system capable of converting a plate number into government knowledge of where a vehicle has been.

An Officer Allegedly Searched His Estranged Wife 717 Times

The most direct answer to anyone who dismisses personal abuse as speculative arrived in Florida in August 2026.

According to an arrest affidavit and the Haines City Police Department, Officer Christopher Goodson queried his estranged wife’s vehicle through the Flock system 717 times between September 1, 2024 and June 30, 2026. Investigators alleged that she was not the subject of a criminal investigation and that her vehicle was not connected to a crime that justified the searches. Goodson reportedly told a supervisor that he had been using the system to determine the location of his children while they were with their mother. The affidavit further alleged that law-enforcement categories including drugs and narcotics, wanted person, motor-vehicle offense and other criminal classifications were entered as reasons for searches. Goodson was arrested on charges of official misconduct and unauthorized computer-system access. The charges remain allegations unless proved.

There is very little abstraction left in that example.

A government employee allegedly wanted to know where his estranged wife’s vehicle was. The government’s surveillance infrastructure could tell him. According to investigators, he asked it hundreds of times.

The incident is especially important because Haines City Police Chief Isaac Jackson did not defend such activity as an unavoidable gray area. The department said that investigative databases may not be used for unauthorized or personal purposes and announced additional auditing measures after the arrest.

That response is appropriate. It also exposes the underlying design problem. The protection depended on discovering the searches after they had occurred.

The woman did not possess a technical mechanism capable of preventing an officer with authorized credentials from turning the state’s crime-fighting infrastructure toward her private life. The audit could establish what happened later. It could not make her movements unknown again.

It Was Not an Isolated Officer

Georgia provides a more disturbing record because the alleged misuse has appeared across ranks and agencies.

In November 2025, the Georgia Bureau of Investigation arrested Braselton Police Chief Michael Steffman on charges that included stalking, harassing communications, violation of oath and multiple counts of misuse of automated license plate recognition systems. The GBI stated that its investigation found that Steffman had misused ALPR systems to harass and stalk multiple individuals. Those charges likewise remain subject to adjudication.

Then, on July 6, 2026, the GBI announced the arrests of five former Albany Police Department officers after an internal audit of the Flock system. According to the GBI, the audit showed that officers had accessed retained Flock license-plate information on multiple occasions for non-law-enforcement purposes. Collectively, the five officers were charged with 21 counts of misuse of license-plate data, in addition to violation-of-oath charges.

The Associated Press reported this week that a Washington Post investigation had identified nearly 50 instances of law-enforcement officers charged or accused of unauthorized license-plate-reader use, many involving current or former romantic partners or family members. AP also reported that the proliferation of such cases was one of the pressures behind Flock’s decision to make abnormal-search auditing and case-linked search controls mandatory rather than optional.

The number matters less than the demonstrated mechanism and the fact of human nature: Technology will be exploited. Count on it.

Before deployment, the hypothetical sounds alarmist: What if an officer used the system to follow his girlfriend? What if somebody tracked an estranged spouse?

After deployment, those sentences no longer begin with “what if.” It’s only a matter of “when”.

This is the predictable insider-threat problem that accompanies every sufficiently powerful information system. Most police officers may never misuse it. That observation does not answer the policy question. Banking rules are not designed on the assumption that most tellers will steal. Medical privacy laws do not disappear because most nurses will never pull a celebrity’s medical chart out of curiosity. Nuclear-security protocols do not depend upon most employees being trustworthy.

The relevant question is what an individual user can do, how quickly misuse is detected, and what independent authorization stands between curiosity and surveillance.

A Nationwide Search Related to an Abortion Investigation

A different kind of problem surfaced in Texas.

Records obtained by the Electronic Frontier Foundation showed that the Johnson County Sheriff’s Office conducted a Flock search across a network exceeding 83,000 cameras. The search-log description was blunt: “had an abortion, search for female.”

The sheriff’s office and Flock later characterized the search publicly as an effort to locate a missing woman and check on her welfare. Subsequent records obtained by EFF included a sworn affidavit from the lead detective describing a death investigation involving a reported abortion. Detectives obtained information concerning the woman’s self-managed abortion, interviewed her, reviewed text messages, and consulted the district attorney’s office about whether she could be charged. Prosecutors advised that state law did not permit charging her for taking the medication involved.

Whether every database query in that investigation violated then-applicable law is a separate question and should not be declared by assertion. The example is important for another reason.

A county sheriff’s office had the technical capability to turn a local investigation into a search across tens of thousands of cameras.

That is the scale problem in concrete form.

The city council that approves several cameras for stolen-car recovery may believe it has bought a local public-safety tool. Network interoperability can make those same cameras components of a search performed by an agency hundreds or thousands of miles away for an investigative purpose the local government never discussed when approving them.

The camera has not moved.

The jurisdiction of the search has.

That is why restrictions on data sharing are not secondary privacy details. They determine who actually possesses surveillance power over the residents of a community.

Then There Is the Problem of the Wrong Person

Intentional misuse is only one danger. The system can be operated with entirely legitimate motives and still place an innocent person into a high-risk police encounter.

Brandon Upchurch’s case in Toledo is among the clearest examples.

In April 2024, a Flock alert led Toledo police to believe that Upchurch’s truck carried a stolen plate. Body-camera footage later showed officers recognizing that the plate information did not match and referring to the reader result as a “mishit” or “misread.” Upchurch was confronted by armed officers, a police dog was released during the encounter, he suffered multiple bites, and he was taken into custody.

Subsequent reporting and court records established the underlying plate discrepancy. The stolen plate was JLL 7362; Upchurch’s plate was JLL 7637. Reporting on the case found that the Flock system had misread a character. Toledo’s internal review resulted in a reprimand for failing to visually verify the plate-reader information before enforcement action, and the incident prompted policy changes governing canine deployment. Upchurch later brought a federal civil-rights action, and Toledo agreed to a $35,000 settlement.

The lesson should not be reduced to “artificial intelligence makes mistakes.” Humans make mistakes as well.

The problem is error amplification.

Before automated surveillance, an incorrect character in a database may remain an incorrect character in a database. Connect that record to an automated system scanning thousands of vehicles and the bad datum acquires reach. Connect it to instantaneous police alerts and it acquires conasequences. Add an officer who assumes the machine has already performed the verification and the transition from database error to armed encounter can occur within minutes.

A similar failure occurred in Plymouth, Minnesota, in June 2026, although the source of the mistake was different and therefore worth distinguishing.

Automotive journalist Joel Feder was driving a manufacturer-owned Range Rover carrying a New Jersey manufacturer’s plate displaying the sequence “34 10 DTM,” with the “10” rendered in smaller characters. Police surrounded the vehicle after Flock alerts associated it with a stolen plate. Plymouth’s police chief later said that a stolen plate had been entered into the National Crime Information Center in California with incomplete characters. The Flock system was matching against that erroneous national record. Feder’s encounter ended without the violence seen in the Toledo case, but four police vehicles converged on him and his wife before officers resolved the mistake.

That case is particularly instructive because blaming “the camera” would be technically inaccurate. The underlying NCIC record was wrong.

But that is exactly why the architecture deserves scrutiny.

A nationwide surveillance network does not merely automate correct information. It automates whatever information it is given.

These Failures Were Predictable Before Flock

None of this began with one company.

Years before Flock reached its current scale, automated license-plate-reader errors were already producing armed encounters.

San Francisco resident Denise Green was stopped in 2009 after an ALPR misread one character of her plate, reading a 7 instead of a 3. Officers confronted her at gunpoint, handcuffed her and searched her vehicle even though the stolen vehicle associated with the erroneous plate was reportedly a gray GMC truck while Green was driving a burgundy Lexus. Her civil case eventually resulted in a $495,000 settlement.

In Aurora, Colorado, another mistaken stolen-vehicle identification led police in 2020 to order Brittney Gilliam and four girls, including children, out of their SUV at gunpoint and place some of them in handcuffs. The vehicle was not stolen. Aurora eventually agreed to a $1.9 million settlement resolving the family’s civil lawsuit.

These older cases matter because they remove another convenient defense. The possibility that automated plate information would produce false police encounters was not unknowable when today’s national systems were built.

The failure mode had already occurred.

The rational response to a known failure mode is engineering and legal redundancy: verify the plate; verify the vehicle; distinguish an automated lead from reasonable suspicion or probable cause; require an auditable investigative purpose; prohibit personal queries; sharply limit sharing; detect repeated searches of the same uninvolved person; and place a judge between government and sufficiently revealing historical searches.

Some states are now writing exactly those safeguards into law.

They are doing so because the empirical question has already been answered.

We do not have to decide whether a nationwide searchable movement database might someday be abused by an officer for personal reasons. Officers have already been arrested for allegedly doing exactly that. We do not have to speculate whether an erroneous plate match might someday cause police to confront an innocent driver as a felony suspect. Innocent drivers have already been held at gunpoint, bitten by a police dog, handcuffed and taken to jail after ALPR-related errors.

Nor does one have to believe that Flock is uniquely malicious, that police generally act in bad faith, or that ALPR technology has no legitimate use.

Those propositions are unnecessary to the argument.

The case for limits is much simpler.

We built a system with extraordinary surveillance power. We now have a record showing the ways that power fails when people misuse it, when databases contain errors, when automated matches are trusted too quickly and when local data become nationally searchable.

Once the failure modes are known, continuing to deploy the capability without corresponding safeguards is no longer an experiment conducted in ignorance.

The Three Lessons

There are three lessons here, and intellectual honesty requires acknowledging both.

First, this is not about whether we should trust the camera, or laws that require warrants. This is about whether the state should be considered to be complicit in handing private information to criminals-in-office who would abuse it. And on this issue, the public will consider the state complicit, regardless of what courts say.

Second, audit logs work. Several of these investigations arose precisely because activity could be reviewed after the fact. That is an argument for mandatory, immutable audit trails, not against them.

Third, an audit that discovers an unauthorized search after somebody has already been tracked is not the same thing as a legal rule preventing an unjustified search in the first place.

A constitutional system normally does not say: search first, document it carefully, and we will find out later whether you had authority. For sufficiently revealing historical or reverse-location searches, that is why warrants exist.

Fifty States, Fifty Very Different Levels of Protection

The Legislative Analysis and Public Policy Association completed a detailed 50-state survey in September 2025. At that time, only 23 states, plus the District of Columbia and U.S. Virgin Islands, had statutes or regulations specifically governing ALPR systems. Those 23 were Alabama, Arkansas, California, Colorado, Florida, Georgia, Idaho, Illinois, Kansas, Maine, Maryland, Minnesota, Montana, Nebraska, New Hampshire, New York, North Carolina, Oklahoma, Rhode Island, Tennessee, Utah, Vermont and Virginia.

Several legislatures acted in 2026, including Washington, Oregon, Kentucky, Connecticut, New Mexico, New Jersey and Indiana. The table below therefore updates the September 2025 national survey with 2026 enactments verified through primary legislative sources as of August 15, 2026.

The rating in the final column is an analytical assessment, not a legal classification. It asks whether a state meaningfully restricts government purposes, retention, querying, sharing and misuse. “Gap” does not mean that citizens have no privacy rights whatsoever. Federal constitutional law, state constitutional protections, general privacy statutes, local ordinances, agency policies and ordinary criminal or civil law may still apply. It means no broad statewide ALPR-specific protection comparable to the stronger regimes was identified in this review.

The table is striking for reasons beyond the number of empty cells.

The states that have seriously confronted this technology are converging on several principles independently: keep ordinary data briefly; limit the reasons an officer may search; record who searched, why and under what case number; restrict downstream sharing; prevent an automated hit from becoming automatic reasonable suspicion; treat sensitive locations differently; impose consequences for intentional misuse; and require judicial authority when the government wants information beyond the ordinary immediate crime-fighting function.

Those are not anti-police principles. They are rules governing a powerful police tool. There is a difference.

Washington Has Come Closest to Writing the Law We Need

Washington’s 2026 statute deserves national attention because it treats ALPR surveillance as a system rather than as a camera.

The legislature begins by recognizing the dramatic expansion of surveillance technology and the need to prevent warrantless monitoring of people’s location and travel from colliding with constitutional protections. It defines ALPR data broadly enough to include location, date, time, photographs, plate numbers and vehicle characteristics. Critically, it also recognizes searchable or analyzed records used to identify, track or reconstruct a person’s or vehicle’s location.

That is conceptually correct. The database is the technology.

Washington then limits law-enforcement use to specified purposes. It prevents an ALPR hit alone from supplying reasonable suspicion without verification or independent grounds. Ordinary records generally disappear after 21 days. Agencies cannot simply sell, rent or purchase the data. Privately held ALPR information requires a probable-cause warrant. Vendors must prevent unauthorized sharing. Agencies must keep detailed audit trails identifying the user, time, query, purpose, offense, case number and cameras involved.

The statute also provides something conspicuously missing from many privacy laws: remedies.

Information knowingly obtained through the prohibited conduct specified by the act can be inadmissible. Intentional unauthorized querying or dissemination can be criminal. Injured persons can seek civil and equitable relief.

A right with no remedy is often an aspiration. Washington supplied some teeth.

Its law is not necessarily perfect. Twenty-one days may still be enough to reconstruct sensitive activity - and does nothing to prevent government complicity in abuses. Enumerated offenses and preservation exceptions require careful implementation. The statute’s effectiveness will depend heavily upon audit quality, enforcement and whether governments comply with the rule’s substance rather than merely its paperwork.

But Washington has answered a foundational question correctly: government surveillance capability should be bounded by law before someone abuses it, not by vendor promises after public controversy.

What a Serious National Model Would Look Like

Congress could legislate here, but states do not need to wait.

The best model can already be assembled from provisions enacted around the country.

First, however, stiff penalties for abuses including mandatory minimum jail time and heavy fines are an absolutely must. Government employees must be put on notice that their actions in the abuse of the system will be prosecuted. Keystrokes must be logged, and potential violations and their possible consequences must be communicated to the government employee users before they hit send. Each and every search.

The other safeguards have precedence. Ordinary non-hit data should expire automatically after no more than seven to 21 days. An agency with probable cause should be able to preserve identified evidence for an investigation or prosecution, but preservation should be particularized rather than turning an exception into indefinite retention of the whole database. Defense counsel must also have a mechanism to preserve potentially exculpatory records once a case exists; privacy should not become an excuse for evidence destruction.

Real-time hot-list functions should remain available for stolen vehicles, missing and endangered persons, serious outstanding warrants and vehicles credibly connected to significant crimes. An algorithmic hit should be treated as an investigative lead, not as magic probable cause. New Hampshire and Washington both recognize this.

Historical person-specific reconstruction should ordinarily require a warrant once the requested search is sufficiently revealing. Reverse-location searches should receive even greater scrutiny because the search begins with a population rather than a suspect. A warrant should identify the place, time, offense, search parameters and justification for examining persons whose records will necessarily be swept into the query.

Sensitive-location protections should become standard. Churches, political events, medical facilities, attorneys’ offices and similar locations can reveal constitutionally protected association or deeply private activity even when a car remains outside on a public street. Chatrie establishes the informational problem; Washington provides one legislative response.

Sharing restrictions must follow the data. A state accomplishes very little by prohibiting its own officers from performing a search while permitting them to ask a neighboring agency, fusion center, private vendor or federal partner to do it for them. Privacy laws need anti-circumvention provisions.

Every query should generate an immutable record: user, agency, time, search terms, location searched, offense, case number, stated purpose, records viewed, records exported and every person or agency to whom data were subsequently disclosed. Those audit records should survive considerably longer than the underlying location data precisely because their purpose is accountability.

Aggregate auditing should be public. Citizens do not need access to the movements of their neighbors, but they should be able to learn how many searches their police department conducted, for what categories of crimes, how frequently other jurisdictions accessed its network, how many violations audits discovered and what happened when employees violated policy.

Vendor contracts should make clear that the public agency controls the information; the vendor may not sell it, repurpose it, use it to construct an independent commercial movement database or silently alter sharing permissions. Washington expressly addresses several of those issues. Oregon likewise imposes vendor obligations and allows civil action for improper vendor use or disclosure.

Intentional misuse should carry consequences that cannot be reduced to retraining. Georgia’s experience demonstrates why.

And deployment itself should occur in public. A city council should know how many cameras it is authorizing, where fixed cameras will generally operate, how the data can be searched, who can access them, which jurisdictions can share them and what the contract actually permits. “The police bought some cameras” is no longer an adequate description of what the government acquired.

It bought access to a database architecture. That architecture should be governed accordingly.

The Clean Constitutional Case Is Still Waiting to Be Built

Civil-liberties lawyers seeking the strongest challenge should resist the temptation to claim that every Flock camera violates the Fourth Amendment.

That claim asks the court to discard too much settled law unnecessarily.

A better plaintiff would have a documented history showing repeated captures across enough locations to reconstruct meaningful movements. Discovery would establish actual historical searches, preferably more than one, along with the audit logs showing who searched, what parameters were used and which other agencies received access.

An especially strong record would show that the system revealed attendance at a religious service, political gathering, medical appointment, legal consultation or another sensitive association. Another strong case would involve a reverse-location search in which police began with everyone who had been near a place and used the database to identify potential suspects.

Then the court could confront an actual search rather than an abstract possibility.

That distinction may also answer one weakness in Schmidt. A plaintiff whose records have merely been collected is asking a court to rule on the surveillance architecture before the government has necessarily used it against him. A plaintiff whose plate history has actually been queried presents the court with the constitutional event in completed form.

The requested rule can then be narrow: Government may use ALPR technology for immediate, specifically authorized public-safety purposes, but sufficiently revealing historical tracking of an identified person’s vehicle, and reverse-location searches that examine the movements of multiple people to discover a suspect, require a warrant supported by probable cause and particularity unless a recognized exigency applies.

Courts do not have to ban the camera. They have to regulate the search. That is a much stronger case.

Public Safety Is Not the Enemy of the Argument

Flock and police departments can point to legitimate successes involving stolen vehicles, missing persons and serious crimes. Flock currently reports that its technology helps agencies locate missing people and solve cases. Those claims deserve empirical scrutiny when used to justify particular deployments, but there is nothing inherently implausible about the underlying proposition that rapid vehicle identification can solve crimes and save time.

Civil libertarians weaken their own argument when they pretend the technology has no legitimate function.

The correct question is one we have encountered repeatedly whenever technology dramatically lowers the cost of government surveillance: Does usefulness eliminate the need for a warrant? It never has.

Telephone wiretaps are useful. Searches of homes are useful. DNA databases are useful. Access to electronic communications is useful. Search warrants themselves exist because useful investigative activity can also be intrusive investigative activity.

The Constitution does not require police to remain technologically primitive. It requires government power to remain legally bounded.

The most important achievement of the emerging state statutes is therefore not a particular retention number. It is their rejection of the proposition that because every individual observation could have been made lawfully, every aggregation and subsequent use of those observations is automatically lawful as well.

That proposition cannot survive indefinitely in a country covered by machine vision.

Criminal Liability Already Exists. It Does Not Answer the Charge Against Government.

There is an important distinction that should not be lost in the debate over Flock and other automated license plate reader systems.

Some of the conduct we are worried about is already criminal.

Washington has now made that explicit. Under its 2026 ALPR law, any person who willfully and intentionally queries, accesses or uses an ALPR system for a purpose not specifically authorized by the statute, or intentionally retains, sells, shares, permits access to or disseminates ALPR or audit-trail data in violation of the law, commits a gross misdemeanor. Washington also provides civil remedies and excludes information knowingly obtained through certain unlawful ALPR practices. A Washington gross misdemeanor for which no different punishment is specified can carry up to 364 days in jail and a fine of up to $5,000.

Georgia had already gone down this road. O.C.G.A. §35-1-22 makes it a misdemeanor of a high and aggravated nature to knowingly request, use, obtain or attempt to obtain law-enforcement license-plate data under false pretenses or for a purpose other than a law-enforcement purpose. The Georgia Attorney General has specifically recognized that offense, and Georgia investigators have actually used the law. In July 2026, for example, the Georgia Bureau of Investigation arrested five former Albany Police Department officers after an internal audit allegedly showed repeated use of retained Flock data for non-law-enforcement purposes. The five collectively faced 21 counts of misuse of license-plate data in addition to violation-of-oath charges.

North Carolina provides another explicit example. Its ALPR statute makes improperly obtaining, accessing, preserving or disclosing data a Class 1 misdemeanor. Maine goes further structurally: its statute generally prohibits ALPR use except for specified governmental and law-enforcement purposes, requires authorized users to act only for the legitimate purposes of their employers, limits ordinary data retention to 21 days, and makes violation of the section a Class E crime.

Even where a state lacks an ALPR-specific criminal statute, existing criminal law can reach some forms of abuse. Florida, for example, makes it a third-degree felony to willfully and knowingly access a computer system without authorization or in a manner that exceeds authorization. Its official-misconduct law can separately criminalize specified forms of intentional misconduct by public servants. Whether either statute applies to a particular ALPR search depends upon the facts and elements of the offense; the point is that a government employee’s badge and password do not necessarily immunize personal misuse of a government database.

Federal criminal law can also enter the picture in sufficiently serious cases, although it should not be invoked casually. 18 U.S.C. §242 criminalizes a willful deprivation of federally protected rights under color of law. The Computer Fraud and Abuse Act, 18 U.S.C. §1030, separately addresses unauthorized access and certain circumstances in which authorized access is exceeded. Neither statute makes every improper database search a federal crime; their elements must actually be proved. They nevertheless illustrate an important point: existing law already recognizes that possessing governmental authority or computer credentials does not create an unlimited right to use them.

So the question is no longer whether society knows how to criminalize a rogue government employee who abuses a database.

We do.

The harder question is whether punishing the rogue employee afterward answers the objection to government having created the surveillance capability in the first place.

It does not.

The public’s use of the word complicit here should be understood carefully. I am not asserting that a legislature, municipality or police department automatically becomes a criminal accomplice under state or federal complicity law whenever an employee unlawfully searches an ALPR database. Criminal accomplice liability has specific elements, including requirements concerning intent that cannot simply be assumed.

The argument is about institutional complicity.

Government purchases the cameras. Government permits the collection. Government authorizes access to the network. Government decides how much location information may be retained. Government determines how many agencies can search it. Government supplies credentials to employees. Government creates, or permits a private contractor to create, the historical movement database.

Then an employee abuses it.

The standard institutional response is that the employee violated policy, exceeded authorization, committed a crime or betrayed the public trust.

Perhaps all of those statements are true. They still leave unanswered the prior question:

Why did government create a system capable of revealing this information about an innocent citizen to that employee in the first place?

That is where criminalizing misuse reaches its conceptual limit.

Washington’s statute illustrates the distinction particularly well. It is a substantial improvement because it expressly criminalizes unauthorized queries and dissemination. But notice what the criminal prohibition necessarily assumes: there is an authorized universe of queries and an unauthorized universe of queries. The criminal law acts at the boundary.

If government itself authorizes mass collection, broad retention and widespread querying, the criminal-misuse statute does not solve that problem. It cannot, because the conduct being challenged has been placed on the “lawful” side of the boundary.

This is why prosecutions of individual officers, while necessary, cannot be the public’s final remedy.

Suppose an officer searches an ex-wife’s vehicle 700 times for personal reasons. Prosecuting that officer may punish the abuse. It may deter another officer. An audit may reveal every keystroke. A civil judgment may compensate the victim.

None of those things can make the victim’s movements private again.

More fundamentally, none answers why information about the movements of an innocent person, who had never been suspected of a crime, was sitting inside a government-accessible historical surveillance system waiting to be searched.

An audit log tells us who opened the door.

The prior question is why government built the room containing everyone’s files.

This distinction becomes even more important as network scale increases. When an officer’s unauthorized conduct consists of opening a personnel file stored by his own department, the state has a legitimate reason to possess the personnel file. When an emergency-room employee improperly opens a patient’s medical record, the hospital plainly had a legitimate reason to possess that patient’s chart.

A mass movement database is different because government’s decision to possess the information is itself part of the controversy.

That is the question criminal prosecution cannot resolve.

The five Albany officers can be prosecuted if the state proves its allegations. Washington can prosecute an employee who deliberately conducts a prohibited search. North Carolina and Maine can punish statutory violations. Federal prosecutors can use federal law when the evidence actually satisfies its elements. Those prosecutions are appropriate exercises of accountability.

But after enough documented cases of personal misuse, government can no longer plausibly treat insider abuse as an unforeseeable event.

That changes the policy question.

When government knows that a capability can be used to stalk spouses, girlfriends, political opponents, neighbors or strangers; knows that unauthorized searches have actually occurred; knows that identification errors can place innocent motorists into armed encounters; and nevertheless continues accumulating searchable movement histories of people suspected of nothing, citizens are entitled to say that responsibility does not end with the employee who typed the query.

The government chose the architecture.

In that civic and institutional sense, enabling foreseeable abuse is participation in the conditions that make the abuse possible.

The answer cannot merely be a warning screen telling the officer that unauthorized use is illegal. It cannot merely be logging every search so investigators can discover the violation later. Those are worthwhile safeguards, but they are safeguards against abuse of a surveillance system. They do not answer whether the surveillance system should possess the information at all.

That distinction also explains why increasingly severe criminal penalties, by themselves, cannot solve the problem.

Make intentional personal misuse a felony. Impose substantial fines. Require permanent audit records. Inform every government employee before every query that unauthorized searching can result in termination and prosecution. Those measures may be entirely justified.

But a ten-year sentence imposed after an officer learns where an estranged spouse spent the night does not restore the privacy that government surrendered when it gave him access to the information.

The public therefore faces a legislative question that prosecution cannot answer.

What information should government be permitted to collect about innocent people in the first place?

If citizens conclude that government should not maintain a searchable historical record of ordinary Americans’ movements without individualized suspicion, then the durable remedy is not another prosecution after another officer is caught.

The remedy is to change the law.

That can mean prohibiting indiscriminate historical ALPR databases while continuing to allow genuinely targeted public-safety uses: real-time comparison against stolen-vehicle lists, missing-child alerts, identified wanted vehicles and particularized investigations supported by appropriate judicial process. It can mean automatic destruction of non-hit observations rather than stockpiling them because they might someday become useful. It can mean prohibiting reverse searches of innocent populations without a warrant - with the highest levels of prior evidence possible. It can mean forbidding interstate and federal access that circumvents the protections enacted by the state whose residents generated the records.

In other words, ban mass surveillance rather than pretending that punishing occasional misuse makes mass surveillance safe.

Washington, Georgia, Maine and North Carolina demonstrate that legislators already recognize unauthorized ALPR use as serious enough to justify criminal punishment.

The next question is harder.

If we know that government employees sometimes become the criminals from whom citizens need protection, why are we designing government systems on the assumption that access to everyone’s movements is acceptable so long as employees promise to use it properly?

The Constitution was never based on that assumption.

Neither should the law governing mass surveillance be.

We Should Write the Law for the Government We Distrust Most

There is an easy way to lose perspective on surveillance powers: evaluate them according to whether one trusts the officials currently using them. That is exactly backward.

Imagine the administration you distrust most. The governor you distrust most. The attorney general you distrust most. The sheriff you distrust most. The federal agency you distrust most.

Now give that institution searchable access to the accumulated movements of vehicles traveling throughout the country, allow agencies to exchange those records across jurisdictional boundaries, and permit searches without prior judicial review.

If the architecture still seems harmless, we disagree about the purpose of constitutional government.

Rights are written for the government that might come next.

Chatrie should change the national debate because the Supreme Court has now rejected a comforting but inadequate answer to location surveillance: that a little bit of retrospective tracking is necessarily harmless merely because the government requested only a short interval. The Court recognized that short-term information can reveal sensitive associations and that selecting a small window from an encompassing historical database does not make the architecture disappear.

Flock is different from Google Location History. A license plate is public. A car is not a cell phone. Current ALPR systems usually provide intermittent vehicle observations rather than an exact continuous record of a human being’s whereabouts. Those distinctions matter, and courts are right to consider them.

They are not the end of the analysis.

The technology is scaling precisely because each additional camera and participating jurisdiction makes the database more useful. As network density increases, the factual premise sustaining earlier cases becomes progressively less reliable: what courts once described as isolated observations can become a reconstructable pattern of movement.

The constitutional line will eventually have to be drawn somewhere.

We should draw it before the technological capability becomes so ordinary that government argues its ubiquity has made surveillance socially acceptable.

Permit the hot-list alert for the stolen car. Find the abducted child. Use the cameras to locate the vehicle credibly connected to a violent crime. Preserve particular evidence when a court authorizes it.

But when government wants to open the historical database and reconstruct where a person has traveled, or search everyone who happened to be somewhere in order to determine whom to investigate, make the government do what the Fourth Amendment has required for centuries when investigative power becomes sufficiently intrusive: State the probable cause. Describe the search. Put it before a judge.

What should not become normal is the proposition that government may collect everybody’s movements first and decide whom to investigate later.

That is the logic of the general search rendered in software.

The surveillance state does not arrive on the day a legislature abolishes the Fourth Amendment. It arrives when technology quietly makes the Fourth Amendment seem inconvenient.

Let’s pass laws that educate would-be abusers on their duties to protect individual liberties and the severe penalties if they ignore the warning.

Or, let’s pass laws banning mass surveillance so we push back completely on the Surveillance State.

This time is precisely when we need them most.

Subscribe now

Share

Leave a comment

Source Notes

1. Flock Safety, “Flock Guardrails Address LPR Privacy Concerns and Police Transparency,” Aug. 13, 2026. Source

2. Flock Safety, “The Price of Free LPR,” company network description. Source

3. U.S. Supreme Court, Chatrie v. United States, No. 25-112 (June 29, 2026). Source

4. U.S. Court of Appeals for the Fifth Circuit, United States v. Porter (Mar. 17, 2026). Source

5. U.S. District Court, Schmidt v. City of Norfolk (Jan. 2026). Source

6. Massachusetts materials discussing Commonwealth v. McCarthy and ALPR surveillance. Source

7. Washington, Chapter 239, Laws of 2026 (ALPR regulation). Source

8. Legislative Analysis and Public Policy Association, Automatic License Plate Recognition Systems: Summary of State Laws (Sept. 2025). Source

9. Georgia Bureau of Investigation, July 6, 2026 ALPR misuse arrests. Source

10. Georgia Bureau of Investigation, July 23, 2026 ALPR misuse arrest. Source

11. Georgia Bureau of Investigation, July 27, 2026 ALPR misuse arrest. Source

12. Georgia Bureau of Investigation, Aug. 7, 2026 ALPR misuse arrest. Source

13. Connecticut Public Act 26-14. Source

14. Kentucky HB 58 / Acts Chapter 71 (2026). Source

15. New Hampshire RSA 261:75-b. Source

16. New Jersey A4070 / P.L.2026, c.4. Source

17. New Mexico SB 40, Driver Privacy and Safety Act (2026). Source

18. Oregon SB 1516 / Chapter 77 (2026). Source

19. Virginia Code §2.2-5517. Source

20. Vermont Statutes, Title 23, Chapter 15. Source

21. Michigan HB 5492 (introduced 2026). Source

22. Washington State Legislature, Engrossed Substitute Senate Bill 6002 (Driver Privacy Act / Chapter 239, Laws of 2026), including the criminal-penalties section making willful unauthorized ALPR querying, access, use, retention, or dissemination a gross misdemeanor. https://lawfilesext.leg.wa.gov/biennium/2025-26/Htm/Bills/Senate%20Passed%20Legislature/6002-S.PL.htm

    Washington RCW Title 10 new chapter on automated license plate reader systems (effective March 30, 2026), including operation limits, retention rules, and criminal/civil remedies. https://app.leg.wa.gov/RCW/default.aspx?cite=10.130.030 (and related sections; full bill text linked above)

    Georgia Code Annotated § 35-1-22 (captured license plate data; penalty), subsection (d)(1) — knowing non-law-enforcement use or access under false pretenses is a misdemeanor of a high and aggravated nature. https://law.justia.com/codes/georgia/title-35/chapter-1/section-35-1-22/

    North Carolina General Statutes § 20-183.33 — Class 1 misdemeanor for obtaining, accessing, preserving, or disclosing ALPR data other than as allowed by Article 3D. https://law.justia.com/codes/north-carolina/chapter-20/article-3d/section-20-183-33/ https://www.ncleg.gov/EnactedLegislation/Statutes/PDF/ByArticle/Chapter_20/Article_3D.pdf

    Maine Revised Statutes Title 29-A § 2117-A (use of automated license plate recognition systems) — Class E crime for violations of the statutory restrictions. https://www.mainelegislature.org/legis/statutes/29-a/title29-Asec2117-A.pdf

    Legislative Analysis and Public Policy Association, Automatic License Plate Recognition Systems: Summary of State Laws (September 2025, with later updates referenced in 2026 analyses) — identifies criminal-penalty provisions in Georgia, Maine, Maryland, North Carolina, Utah, Virginia, and others. https://legislativeanalysis.org/wp-content/uploads/2025/09/Automatic-License-Plate-Recognition-Systems-Summary-of-State-Laws.pdf

    Georgia Bureau of Investigation press releases (July–August 2026) documenting enforcement of O.C.G.A. § 35-1-22 against officers/employees for non-law-enforcement ALPR access (Albany, Conyers, and related cases). https://gbi.georgia.gov/press-releases/2026-07-06/gbi-arrests-five-former-albany-police-department-officers-misuse-license https://gbi.georgia.gov/press-releases/2026-07-23/gbi-arrests-conyers-police-department-employee-misuse-license-plate

Another question deserves attention: What did the extraordinary demands of World War II do to the structure of modern states, and how much of that emergency architecture survived into the Cold War?

World War II required governments to mobilize society on a scale previously difficult to imagine. Industrial production, transportation, scientific research, intelligence, communications, food, labor, propaganda, taxation, censorship, military planning, and technological development were pulled into national war-making systems. In the United States, a constitutional republic created enormous administrative and military capacities because defeating the Axis powers required them. In the Soviet Union, an already centralized revolutionary state intensified command authority still further under the pressures of invasion and total war.

The war ended. Much of the machinery did not. That, in itself, is a lesson: Do not build structure in government that you do not expect to be inherited and used against your political ideals.

That matters because the Cold War was not simply a contest between Washington and Moscow. It became a contest conducted through enormous permanent institutions whose continued existence depended upon identifying, anticipating, managing, and sometimes secretly confronting threats that ordinary citizens could neither independently verify nor directly supervise.

In the communist bloc, the concentration of political and economic authority was explicit. Communist governments generally asserted that central planning and disciplined political leadership were necessary to defend the revolution, reorganize economic life, resist foreign interference, and suppress forces considered counterrevolutionary. The result was political power concentrated within party structures, security services, military institutions, and state bureaucracies that sharply constrained independent political competition and direct popular control.

The American path was profoundly different in law, ideology, political culture, and degree, yet it produced its own form of centralization.

The United States did not become a communist command state. It retained elections, competing political parties, private property, constitutional institutions, courts, an independent press, and substantial civil society. Yet after World War II it also developed something unprecedented in American history: a large, permanent national-security apparatus operating continuously in peacetime.

The National Security Act of 1947 created the institutional framework from which the modern Department of Defense, National Security Council, and Central Intelligence Agency emerged. Nuclear weapons required extraordinary secrecy. Global military commitments required standing forces and overseas bases. Intelligence operations required compartmentalization. The Soviet threat encouraged continuity of emergency planning. NATO internationalized military obligations. Proxy conflicts blurred the boundary between war and peace. Presidents acquired increasingly powerful national-security instruments whose operations often occurred far beyond the ordinary visibility of Congress, courts, journalists, and voters.

What emerged might be called the Agency State: not government by a single agency, but a political condition in which permanent bureaucratic institutions possess specialized knowledge, classified information, rule-making operational continuity, and decision-making capabilities that elected officials and ordinary citizens cannot easily reproduce or independently evaluate.

This produces one of the Cold War’s most enduring paradoxes.

The communist model centralized power openly in the name of revolutionary necessity and state planning. The American national-security model centralized important forms of power indirectly, through expertise, secrecy, delegated authority, emergency doctrine, intelligence operations, military commitments, and bureaucratic permanence.

These systems were not morally or politically equivalent. Their legal protections, levels of coercion, economic structures, treatment of dissent, and opportunities for political opposition differed enormously. Yet viewed as systems of political organization, they raise a common problem: when institutions acquire extraordinary authority in response to extraordinary threats, how easily does that authority return to the public once the emergency subsides or becomes permanent?

That question becomes especially important because the Cold War was rarely cold.

Korea and Vietnam became devastating wars. Afghanistan became a battlefield between Soviet forces and Western-supported resistance. Latin America became an arena for revolutionary movements, counterinsurgencies, coups, covert interventions, intelligence cooperation, narcotics networks, and political violence. Decolonization movements were repeatedly interpreted through the lens of superpower competition. Governments that might otherwise have been treated as local actors were classified as pieces on a global ideological chessboard.

Operations and programs such as Gladio, Condor, the Phoenix Program, Paperclip, MK-Ultra, and the proposed Operation Northwoods plan compel us to confront difficult questions about what governments are willing to consider or undertake when national security is invoked. They also force an analytical distinction that is too often lost in contemporary political argument: discovering that governments concealed, manipulated, or abused power in some circumstances does not establish that every allegation of covert activity is true. Skepticism toward official narratives must be matched by skepticism toward unsupported counter-narratives.

That is where serious historical inquiry becomes most valuable.

The Cold War cannot be understood through propaganda from either side. Moscow had its mythology. Washington had its mythology. Communist governments portrayed Western democracy as little more than capitalist domination while minimizing their own repression. Western governments portrayed themselves as defenders of freedom while sometimes supporting authoritarian regimes, conducting covert operations both foreign and domestic, shaping public opinion, or tolerating actions abroad that would have been difficult to defend under ordinary democratic scrutiny at home.

Meanwhile, ordinary people lived inside these systems.

Eastern Europeans navigated states whose governments frequently constrained political speech, association, movement, religion, and economic independence. Americans lived under a constitutional order that remained vastly freer while nevertheless experiencing loyalty investigations, classified government programs, expanding surveillance capacities, military conscription, propaganda, domestic intelligence operations, and repeated arguments that national survival required limiting public knowledge of what their government was doing.

Both systems therefore illuminate a problem larger than the Cold War itself: centralized power justifies and expresses itself differently depending upon the political system in which it develops.

One system may invoke equality, revolution, solidarity, or protection from counterrevolution. Another may invoke security, expertise, continuity, deterrence, or protection from foreign enemies. The vocabulary changes. The institutional temptation does not.

This is one reason the Cold War still matters thirty-five years after the Soviet Union disappeared.

The institutions built during that struggle did not simply vanish in 1991. Intelligence agencies, NATO remained, nuclear command systems remained, the national-security classification system remained, defense contracting remained. The techniques of covert action, information management, surveillance, psychological operations, strategic communications, and threat assessment continued evolving. The administrative state remained. Russia inherited much of the Soviet security apparatus, while other former communist states struggled with the political and institutional inheritance of centralized rule.

The Cold War therefore deserves to be studied not merely as a finished historical period but as the construction site of much of the political world we inhabit today.

That is the premise behind The Cold War: Lingering Questions, a 19-lesson course taught by Brad Miller, beginning Thursday, September 24, 2026 at 7:00 p.m. ET. IPAK-EDU.ONLINE members receive and automatic 30% off the list price.

SEE THE NEW COURSE

The course examines the conditions that created the Cold War, the strategic reasoning of Washington and Moscow, the rise of the Soviet Union as America’s postwar rival, Korea, Vietnam, Afghanistan, Latin America’s dirty wars, espionage, the nuclear and space races, the CIA, National Security Council and Department of Defense, the creation of the United Nations and NATO, Eastern Bloc experiences, the Non-Aligned Movement, propaganda, covert operations, constitutional limits, government secrecy, and the enduring tension between national-security requirements and democratic accountability.

Most importantly, it approaches controversial history with a discipline badly needed in our present age: distinguishing documented fact from reasonable inference, and reasonable inference from speculation.

The most interesting question about the Cold War may therefore no longer be simply, Who won?

It may be this:

What kinds of states did forty-five years of permanent geopolitical emergency create—and how much of that architecture are we still living inside?

Sign up and explore The Cold War: Lingering Questions and enroll here:

SIGN ME UP!

PS: REMEMBER - IPAK-EDU.ONLINE members receive and automatic 30% off the list price of all of our courses and webinars.

On October 2, 2025, Gallup reported that only 28% of Americans had a great deal or fair amount of confidence in newspapers, television and radio to report the news “fully, accurately and fairly.” It was the first reading below 30% in Gallup’s series. Republican confidence had fallen to 8%, independent confidence stood at 27%, and even among Democrats the figure was 51%, tied for that group’s lowest reading in the series (Brenan, 2025).

The Reuters Institute reached a similar finding using a different question and therefore a different measure. Its 2026 Digital News Report found that 25% of Americans said they trusted the news most of the time, down five percentage points in a year and the lowest U.S. figure Reuters had recorded since it began tracking the measure in 2015 (Reuters Institute, 2026).

Those measures should not be treated as interchangeable, but they establish the condition in which Americans now consume news: confidence is low, and it is profoundly conditioned by political affiliation. Pew Research Center reported in February 2026 that 57% of U.S. adults had little or no confidence that journalists act in the public’s best interests. Among Republicans and Republican-leaning independents, only 25% expressed a great deal or fair amount of confidence; among Democrats and Democratic leaners, 61% did (Eddy, 2026).

The usual response to this problem has been to tell readers more about which outlets they should trust.

I think we have spent enough time doing that.

The more useful question is whether we can measure what happens to a story as it passes through different newsrooms.

That is what we built IPAK.news to do.

The Unit of Analysis Has Been Wrong

Consider one of the best recent maps of the American news ecosystem. In June 2025, Pew Research Center examined the use and trust of 30 major news sources and found striking partisan separation. Fifty-seven percent of Republicans said they regularly obtained news from Fox News, while Democrats were much more likely to use CNN, NBC, ABC, CBS, MSNBC, NPR, PBS and The New York Times. Republicans and Democrats also differed sharply in which of those organizations they trusted (Pew Research Center, 2025).

That is important information about audiences.

It is not a measurement of the reporting.

Pew makes the distinction explicitly. Its analysis of the political composition of each outlet’s audience does not categorize the political lean of the outlet’s content.

That distinction should be obvious, yet much of the public discussion of media bias quietly crosses it. We learn that an outlet has a predominantly liberal audience or conservative audience, receives a left or right rating, or is trusted disproportionately by one political party, and then treat that classification as though it describes every article the organization publishes.

It cannot.

The unit of exposure for the reader is usually the story, and the relevant comparison is often narrower still: how multiple organizations covered the same subject during the same period.

An outlet can report one subject with relatively little interpretive language and another with much more. Two outlets normally placed on opposite camps of the political spectrum can frame a particular event similarly. Publications generally grouped together can differ sharply among themselves. A reporter can write a careful article for an organization whose editorial page is highly ideological. Another article can contain almost entirely accurate factual assertions while producing a strong directional impression through selection, emphasis, language, sourcing and context.

An outlet label cannot resolve any of this.

It was never designed to.

Framing Is Real, and It Is Not the Same as Factual Accuracy

The distinction matters because framing has a substantial research literature behind it.

Chong and Druckman experimentally examined how competing frames influence political opinions and found that framing effects depended on the strength of the frames and the competitive information environment. Competition altered the effects; it did not simply make them disappear (Chong & Druckman, 2007).

There is another complication that any serious system for examining media bias has to acknowledge. The reader also brings biases to the encounter. Vallone, Ross and Lepper’s classic work on the “hostile media phenomenon” showed that opposed partisans could encounter the same coverage and perceive it as biased against their own side (Vallone et al., 1985).

That means asking readers whether an article “seems biased” cannot solve the measurement problem. Perception of bias and properties of the article are related questions, but they are not identical.

IPAK.news therefore does something deliberately narrower.

It measures framing.

It does not score truth.

It does not determine which political faction is correct.

It does not infer that an article with stronger framing is factually wrong, or that an article with weaker framing is factually right.

That limitation is part of the design because collapsing factual accuracy, ideology, framing, credibility and reader agreement into a single “bias score” would simply create another opaque judgment for readers to accept.

We wanted an instrument instead.

Seven Things You Can Actually Look For

When a user searches a subject on IPAK.news, the system gathers recent coverage from a fixed roster of outlets assigned to left and right comparison groups and scores the on-topic articles using seven separate measures.

The first is spin: how much an article interprets and characterizes events rather than simply reporting what occurred. The second, fronting, examines the degree to which reporting depends on unnamed sources rather than people attaching their names to consequential claims. Valence measures how warm or hostile the language is toward the central person or institution. Loaded language examines the amount of emotionally charged or judgment-carrying wording doing work that plain description could otherwise do.

The remaining measures capture other forms of editorial construction. Speculation examines reliance on assertions about motives, intentions or future events rather than established information. The episodic-to-thematic measure asks whether an event is reported principally as an incident or positioned as evidence of a larger pattern. Focus examines whether the article centers on the substance of the issue or shifts attention toward political contest, maneuvering and personalities.

None of those variables proves dishonesty.

That is precisely why they are separated.

A highly thematic article may be excellent journalism if the broader pattern is well supported. Anonymous sources can provide indispensable information. Interpretation can help a reader understand an extraordinarily complicated event. Strong negative valence may be warranted by the underlying record.

The measurement tells you that the characteristic is present and gives you something much more useful to investigate than the generic accusation that an article is “biased.”

Where the Two Camps Actually Separate

IPAK.news then compares the distributions of those measures across the two outlet groups.

This is where the exercise becomes considerably more interesting than assigning permanent ideological labels to newspapers.

For a given topic, left- and right-group articles may be nearly aligned on six of the seven measures and separate sharply on one. They may diverge in valence but remain similar in speculation. They may use comparable language while differing substantially in whether the event is presented as a single episode or evidence of a larger pattern. One group may focus heavily on political maneuvering while the other spends more of its coverage on the substance at issue.

Or they may align.

That result matters too.

IPAK.news is capable of returning ALIGNED, LEANS, DIVERGES, or insufficient-coverage findings. The engine uses a prespecified separation rule rather than deciding after seeing the answer whether a difference “looks large.” Its current framing engine uses a separation gate of δ = 12 points.

That deserves a methodological qualification because I do not want IPAK.news making the same category errors I object to elsewhere. DIVERGES is descriptive, not inferential. It describes separation among the articles gathered and scored in that search window. It is not a p-value, a confidence interval, proof that every publication on one side behaves similarly, or an estimate of some hypothetical population of all American journalism.

The readout also shows the spread of the scored articles within each group. That matters because two camps can have different means while containing substantial internal variation, and an average can conceal exactly the kind of heterogeneity a reader ought to know exists.

This is a measuring instrument. Its scope and limitations should remain visible.

There Is Another Kind of Bias: Silence

A story can also be framed by whether it receives coverage at all.

That problem cannot be reduced to language analysis because there is no article to score. IPAK.news therefore makes search asymmetry visible rather than silently discarding it. Its deeper search attempts to gather substantial coverage from both outlet groups; when one side remains below the target after the available searches have been exhausted, the system can expose that imbalance as a possible silence signal.

Again, that is not proof of ideological suppression. There may be mundane explanations for asymmetric coverage: timing, editorial specialization, access, indexing, geography, or simply different assessments of newsworthiness.

But absence should not be converted into invisible, missing data if both camps are viewing reality through a distorted lens.

If ten organizations associated with one political audience are covering a development and almost none of the comparison outlets are, the reader deserves to know that before being handed a symmetrical-looking answer.

In a future release, we will track the communities’ findings and plot them over time to provide an objective measure of the dynamics of how polarized the media is on a particular topic. That will be interesting - and you will have been part of it. Your participation truly matters.

“Debias This News” Is A New Media Experiment - And You Are the Author

After examining the framing readout, the reader can select Debias This News.

The result is a zero-valence brief intended to remove as much of the measured framing apparatus as possible and return the subject to its factual core.

You are the author, in part, because your search placed the debiasing lens on the topic. And you are not limited to the debiased news summary.

You can read the source articles. Examine the framing scores. Look at the variation within each side. See where the groups align and where they separate. Then read the framing-reduced brief and go back to the originals.

The comparison is the product.

Once you see that two articles can describe substantially the same event while one leads you toward motive, conflict and moral judgment and another toward mechanism, chronology and consequence, the phrase “media bias” becomes less useful as an accusation and more useful as a set of questions.

What was added?

What was omitted?

Who was allowed to characterize the event?

Was the central claim established or inferred?

Was one incident presented as evidence of a pattern?

How much of my reaction came from the facts, how much came from their presentation, and how much did I bring with me before I opened the article?

Those questions put the reader back into the process.

This is my best attempt to allow people to see the facts in the news first, and separate framing and bias so they can make up their own minds.

Best of all: You can publish the unbiased version on your own My Debiased News channel via your ipakedu.online account and share it with the world.

NB: The What We Could Not Resolve section is to be read as “more information is needed”, as is more investigative journalism should be done, not that the facts of the story are unresolved. As in science, more research is often needed to confirm or validate specific conjectures. In this way, this tool provides a uniquely Popperian way of reading the news.

Here’s an example:

Try it Out!

IPAK.news Does Not Need You to Trust IPAK.news - You Are Part of the Process

The collapse of trust in American journalism is usually presented as a problem requiring restoration of trust. There may be institutions deserving of more trust than they currently receive, and others deserving of less. A survey of public confidence cannot adjudicate that question because confidence and trustworthiness are different variables.

IPAK.news takes a different approach.

You should be able to inspect the evidence behind the representation.

That standard should apply to CNN and Fox News, to The New York Times and The Daily Wire, to NPR and the New York Post, and it should apply to us.

If IPAK.news says two groups diverge, you should be able to see the dimension on which they diverged, the direction of that difference, the articles underlying the comparison, the within-group spread, the size of the corpus and the rule that produced the label. If the evidence is inadequate, the system should say so.

The proposition behind IPAK.news is therefore narrower than “the news is biased,” and much more useful.

Americans already know that different news organizations can present the same controversy differently. Research has established that framing can influence how people evaluate political questions, while research on hostile-media perception reminds us that readers themselves are imperfect judges of bias. Current surveys establish that partisan news ecosystems and collapsing confidence are real. What has been missing for the ordinary reader is a practical way to inspect how the coverage itself changes across those ecosystems.

That is the problem IPAK.news was built to address.

Search a subject you know well.

Better yet, search one on which you already have a strong opinion.

Then look at what your preferred news sources are doing to the story.

And look at what the other side is doing too.

And then tell the world what you found.

Help us de-bias the news. We have to pay for tokens, etc so there is a subscription. But you get access to your own, and the community’s collectively debiased news.

The revolution will be posted.

I'm in!

Subscribe now

Share

Leave a comment

Sources

Brenan, M. (2025). Trust in Media at New Low of 28% in U.S. Gallup, October 2, 2025.

Chong, D., & Druckman, J. N. (2007). Framing Public Opinion in Competitive Democracies. American Political Science Review, 101(4), 637–655.

Eddy, K. (2026). Majority of Americans Express Low Confidence in Journalists to Act in Public’s Best Interests. Pew Research Center, February 11, 2026.

Pew Research Center. (2025). The Political Gap in Americans’ News Sources. June 10, 2025.

Reuters Institute for the Study of Journalism. (2026). Digital News Report 2026: United States.

Vallone, R. P., Ross, L., & Lepper, M. R. (1985). The Hostile Media Phenomenon: Biased Perception and Perceptions of Media Bias in Coverage of the Beirut Massacre. Journal of Personality and Social Psychology, 49(3), 577–585.

IPAK, IPAK-EDU and Popular Rationalism are not affiliated with this research.

Call 833-JOIN-NIH or email ccopr@nih.gov
https://go.nih.gov/BPuGtKj

James Lyons-Weiler, PhD
Popular Rationalism

On March 16, 2026, U.S. District Judge Brian E. Murphy temporarily stopped the Trump administration’s first major restructuring of childhood vaccine policy. His order stayed the January memorandum revising the CDC childhood immunization schedule, stayed the appointments of thirteen members of the reconstituted Advisory Committee on Immunization Practices, and stayed every vote taken by that ACIP. The court concluded at the preliminary-injunction stage that the plaintiffs were likely to succeed in showing that the administration’s reconstitution of ACIP and the January schedule changes violated the Administrative Procedure Act.

The practical consequence was substantial. The administration had changed the federal vaccine schedule and changed the body historically used to advise CDC about that schedule; Judge Murphy put both on hold. The administration appealed, and the litigation remains active, with the underlying case currently stayed pending appeal.

President Trump has now answered that judicial intervention directly.

The August 10 executive order expressly says that implementation of the administration’s previous childhood-vaccine directives “has been delayed due to litigation over the composition of the Advisory Committee on Immunization Practices and separate updates to the Federal vaccine schedule.” It then declares the administration’s three-tier “Gold Standard Childhood Vaccine Recommendations” as federal policy and orders executive departments and agencies to take appropriate steps to advance them to the fullest extent allowed by law.

The legal distinction matters. Trump has not vacated Judge Murphy’s injunction; an executive order cannot reverse a federal court order. What he has done is attempt to reverse its policy effect by moving the administration’s vaccine-policy commitment above the institutional chokepoint created when the court froze both the January schedule and the reconstituted ACIP.

That is the proper context for understanding what happened today.

And the schedule itself may eventually prove to be the less consequential part of the order.

The January Reset Is Back as Presidential Policy

When the revised schedule appeared in January, I called it necessary, not radical. The important feature was routinely misrepresented. The federal government was not prohibiting vaccines or removing them from medicine. It was rejecting the assumption that every vaccine appearing on the federal childhood schedule necessarily belonged in the same population-wide recommendation category.

The January HHS assessment recommended maintaining eleven disease targets in the category recommended for all children, while shifting other immunizations toward high-risk recommendations and shared clinical decision-making. It explicitly said that no vaccine should be moved into a “not recommended” category and that vaccines remaining on the schedule should remain available and covered without cost sharing. I described the reform at the time as a movement away from schedule maximalism and toward individualized benefit-risk assessment, informed consent and international comparison.

Murphy’s March injunction blocked the January memorandum and effectively disabled the administration’s reconstituted ACIP while the case proceeded. The May 29 executive order tried to continue the reform through the conventional institutional pathway: CDC and ACIP were instructed to review the HHS assessment and update the schedule where permitted by law. That meant the administration was still depending heavily upon the very ACIP mechanism caught inside the litigation.

Today’s order changes that posture.

Trump now declares as United States policy three distinct recommendation categories. Vaccination against measles, mumps, rubella, diphtheria, tetanus, pertussis, polio, Haemophilus influenzae type B, pneumococcal disease, HPV and varicella remains recommended for all children. RSV monoclonal antibodies, hepatitis A, hepatitis B, meningococcal B, meningococcal ACWY and dengue are designated for certain high-risk groups or populations, while hepatitis A, hepatitis B, rotavirus, meningococcal disease, influenza and COVID-19 may fall under shared clinical decision-making. Some immunizations can therefore appear in more than one category depending upon individual risk.

This distinction will undoubtedly disappear from some headlines. It should not disappear here.

Trump did not ban the childhood vaccines moved out of universal recommendation.

He changed the presumption under which they are offered.

That is exactly what risk stratification is supposed to do. A recommendation for every child makes a population-level claim about benefit and risk. A recommendation for a high-risk child makes a different claim. Shared clinical decision-making recognizes that the correct decision can depend upon age, exposure, disease prevalence, prior immunity, underlying conditions and other characteristics of the individual child. Converting one category into another is therefore a substantive change, but describing it as simply “removing vaccines” obscures what the policy actually does. The January assessment itself proposed keeping every vaccine available while differentiating the strength and target population of the recommendation.

The operative principle has shifted from one schedule as the presumptive answer for everyone toward a layered schedule in which some recommendations remain universal and others require an individualized reason.

That is a meaningful restoration of clinical judgment.

The Larger Change Is Individual Liberty Across the States

The most consequential language in today’s order may be Section 4, because the administration has now connected childhood vaccination policy directly to federal enforcement of parental authority, religious liberty, disability accommodation and equal protection.

Trump directs the Attorney General to take appropriate measures to further meritorious legal actions challenging state laws that conflict with constitutional or federal statutory obligations in those areas, including obligations to provide religious and medical exemptions from childhood immunization requirements where federal law requires them. He then separately orders the Departments of Justice, Education, and Health and Human Services to ensure that their contractors and grantees—including states and localities—comply with those constitutional and federal statutory obligations.

This is a major national policy win for individual liberty.

Subscribe now

It does not federalize every school-vaccine statute, and it does not automatically manufacture a religious exemption in every jurisdiction. The operative language specifically says “to the extent applicable under Federal law,” and the order as a whole must be implemented consistently with existing law. State police powers have not disappeared.

But neither have the Constitution and federal civil-rights statutes.

For years, the practical vaccine-policy question in many states has been framed as though a legislature or health department could resolve the liberty question simply by calling a medical intervention a condition of school attendance. Today’s order says the federal government will examine the other half of that equation: whether the way a state imposes that condition violates religious freedom, disability protections, equal protection or other federal guarantees.

The national landscape makes the significance apparent. Every state and the District of Columbia imposes some vaccination requirements for school attendance. According to the National Conference of State Legislatures’ May 2026 review, twenty-nine states and Washington, D.C., specifically allow religious exemptions; another sixteen allow exemptions for religious or personal reasons; and four states—California, Connecticut, Maine and New York—provide no non-medical exemption at all. Mississippi now allows a religious exemption because a federal district court ordered the state to do so in 2023.

That Mississippi history alone demonstrates why Section 4 is not symbolic language. Federal religious-liberty litigation can change the operating rules of a state vaccination system.

The administration had already begun moving in this direction. In September 2025, HHS’s Office for Civil Rights notified participants in the Vaccines for Children program that providers were required to respect applicable state religious and conscience protections. HHS noted that federal operational funding reached 61 state, local and territorial immunization programs.

The August 10 order broadens the institutional force behind that principle considerably. This is now a DOJ matter, an Education Department matter and an HHS matter. It encompasses litigation, federal grants and contracting relationships, and it expressly identifies states and localities as entities whose compliance must be examined.

Last month I wrote about Gov. Josh Shapiro’s decision to place parental vaccine choice inside his conception of “real freedom.” The point was larger than one governor or one political party: bodily autonomy becomes incoherent when it is treated as a fundamental principle in one medical context and an inconvenience in another.

The August 10 order now puts the federal executive branch on record: parental authority and religious liberty are part of childhood-vaccine policy.

They are no longer external complaints to be considered after public-health policy has already been written.

The Scientific Reform Goes Far Beyond Counting Vaccines

There is another reason the order deserves more careful treatment than a headline announcing that Trump “cut the vaccine schedule.”

The January HHS assessment by Tracy Beth Høeg and Martin Kulldorff did something that American vaccine policy should have been doing routinely: it treated the schedule itself as an object of scientific investigation.

The assessment concluded that the United States was an international outlier in both the number of disease targets and doses routinely recommended for children. It compared U.S. policy with peer developed nations, noted that most of those countries maintain high uptake without the American style of school-entry mandates, and proposed distinguishing consensus vaccines from vaccines whose population-wide recommendation was not shared internationally.

But its more important contribution may be the research program it demanded.

Høeg and Kulldorff called for randomized research where ethically feasible on individual vaccines, combinations and schedules, including designs that randomize the timing of vaccination. They called for longer-term observational research, better evaluation of interactions among vaccines, improved safety surveillance, and ongoing calculation of numbers needed to vaccinate and numbers needed to harm separately for low- and high-risk children.

This is the scientific core that should survive every political administration.

A vaccine schedule is a multi-exposure intervention delivered repeatedly during a rapidly changing period of human development. Evidence that Product A has an acceptable benefit-risk profile under one trial design does not automatically answer every question about Product A administered at Age X with Products B, C and D, followed by another group of exposures at Age Y. Conversely, uncertainty about those interactions does not prove that the interactions cause harm.

The rational answer is to study them.

Today’s executive order converts that research argument into an instruction to HHS. The Task Force on Safer Childhood Vaccines has 90 days to present plans for assessing the ideal timing and sequencing of core childhood vaccines; developing alternatives to aluminum adjuvants and performing comparative safety and efficacy studies; continuously evaluating vaccine-specific risk-benefit profiles using American and international data; improving safety monitoring, transparency and research; and expanding availability of single-antigen vaccines, beginning with measles, mumps and rubella.

That aluminum provision deserves particular attention. On July 2, I documented how limited the U.S. market is when parents or physicians seek aluminum-free alternatives for several pediatric vaccine categories. I argued that the appropriate response was neither panic nor institutional denial, but engineering: develop alternatives, compare them properly and give physicians and families better product-level choices.

The President has now directed HHS to develop alternative adjuvants and conduct comparative safety and efficacy studies.

That is exactly the kind of question regulatory science should be willing to answer empirically.

MMR, Vaccine Spacing, and the Burden of Doing the Research Properly

The order goes further than the January assessment in one important respect.

It states that the combined MMR vaccine should eventually be available as separate measles, mumps and rubella vaccines and says the diseases should be administered as separate single-disease shots once those products are domestically available. It also says that, to the maximum extent feasible, childhood immunizations should be administered at separate medical visits. At the same time, the order instructs HHS to guarantee continued availability of combination vaccines while it works with industry and other countries to create single-vaccine options.

Here the evidentiary distinction needs to remain exact.

The January HHS assessment explicitly stated that it did not evaluate the timing or order of vaccines, apart from HPV dosing. It did, however, identify timing, interactions and schedule design as important unanswered research questions and recommend randomized timing studies.

Therefore, the separate-visit provision should be understood as a policy hypothesis that the administration has now also undertaken an obligation to test, rather than as a result already established by the January assessment.

That is not a reason to avoid the experiment. It is the reason to conduct it correctly.

The studies should measure more than short-term adverse events. They should measure infection and severe disease, completion rates, delayed or missed immunizations, adverse-event profiles, interaction effects, age-specific effects and relevant clinical subgroups. If separating particular vaccines produces better net outcomes, the evidence should show it. If particular combinations or same-visit administrations perform better, that should be allowed to show as well.

A gold-standard-science policy has to be falsifiable even when the hypothesis under examination is favored by the people who ordered the research.

There is also an important choice component here that is being missed. The order directs HHS to create single-vaccine options while guaranteeing continued availability of combination vaccines. A parent who prefers a monovalent option would gain something that presently is not generally available for MMR in the United States; a parent and physician who conclude that the combination product is preferable would not see that product ordered off the market.

That distinction is consistent with the larger architecture of the order: increase the number of decisions available to families while making federal recommendations more specific about who is expected to benefit.

The Strongest Objections Are Testable

Critics of the new policy have raised a serious practical objection: requiring or encouraging more separate visits can increase the opportunity for missed doses and leave children susceptible to infection for longer periods. Major medical organizations also continue to support simultaneous administration for many routine vaccines. AP and Reuters reported immediate opposition to the order on those grounds.

That objection should be measured, not dismissed.

So should the opposite concern: whether administering multiple immunological exposures simultaneously or in close succession produces interactions or subgroup-specific risks that existing studies have insufficient power or follow-up to detect. The HHS assessment specifically identifies limited research on interaction effects and calls for schedule-level studies.

The two questions are not mutually exclusive. A separated schedule might reduce one category of risk while increasing another. A combination product might improve adherence while creating an adverse-event profile different from its component products. Different answers may emerge for different vaccines, ages and risk groups.

That is why the correct scientific unit is not ideology. It is the specific comparison.

This is also where proponents of the reform should insist that the administration hold itself to the standard it is demanding from everyone else. If HHS commissions timing studies, the endpoints and analysis plans should be specified prospectively. Relevant adverse outcomes should not disappear behind composite endpoints. Safety studies should be large enough and long enough to address the events they claim to evaluate. Subgroup analyses should be powered when subgroup heterogeneity is part of the clinical question. The data and protocols should be available for independent inspection.

The goal cannot be to manufacture studies that vindicate a predetermined alternative schedule.

The goal has to be to find the better schedule.

Share

What Changed Today

In January, I argued that the schedule reset was a signal that American public health was moving away from maximalism and back toward informed consent, individual benefit-risk assessment and scientific humility.

Judge Murphy’s March order interrupted that experiment by freezing the January memorandum, the reconstituted ACIP and every vote that committee had taken. The administration’s May response still attempted to work through ACIP. The August 10 order explicitly identifies that litigation as the reason further presidential action became necessary.

Trump has now made the substance of the vaccine reset executive-branch policy while opening a second and potentially more important front.

Across the states, the federal government has been instructed to take religious liberty, parental authority, disability accommodation and equal protection seriously enough to enforce them. DOJ can support meritorious challenges to state laws. DOJ, Education and HHS are directed to police compliance among grantees and contractors, expressly including states and localities. The administration simultaneously preserves access to vaccines, distinguishes universal recommendations from individualized ones, commissions studies of timing and sequencing, seeks single-antigen alternatives, orders work on alternatives to aluminum adjuvants, and requires continuing risk-benefit evaluation.

That is a much larger policy than “Trump reduced the number of childhood vaccines.”

For families concerned about medical autonomy, it is a major national victory because the federal government has altered the default relationship between the citizen and the vaccine-policy apparatus. The parent is no longer treated in the federal policy itself merely as the endpoint of a compliance program. The parent and physician are expressly restored to the decision-making process for a substantial portion of the schedule, while states have been put on notice that vaccination laws do not exist outside the boundaries imposed by constitutional and federal protections.

For science, the standard has changed as well. Population-wide recommendations must be capable of defending their population-wide benefit-risk claims. Schedule assumptions are becoming research questions. Formulation choices are becoming engineering questions. Long-term safety and interaction effects are being placed explicitly on the federal research agenda.

None of that proves in advance what every study will find, and it should not.

That is the point.

The March court order effectively undid the administration’s first operating mechanism for the vaccine reset. The August 10 order does not pretend the court never acted. It responds by putting the policy itself on firmer presidential footing, broadening it beyond ACIP, and adding an enforcement architecture for individual liberty that reaches into the states.

The injunction remains to be litigated.

The vaccine reset no longer waits for it.

Subscribe now

Share

Leave a comment

Sources

1. White House, August 10, 2026. Delivering Gold Standard Childhood Vaccine Recommendations for Americans — operative executive order establishing the three recommendation categories, research directives and Section 4 federal enforcement provisions.

2. U.S. District Court for the District of Massachusetts, March 16, 2026. American Academy of Pediatrics et al. v. Kennedy et al. — Judge Brian E. Murphy’s order staying the January schedule memorandum, thirteen ACIP appointments and all votes of the stayed ACIP.

3. White House, May 29, 2026. Executive Order 14407 — directed CDC and ACIP to review the HHS assessment and update the schedule within lawful authority.

4. Høeg TB, Kulldorff M. January 2, 2026. Assessment of the U.S. Childhood and Adolescent Immunization Schedule Compared to Other Countries — HHS assessment supporting risk-stratified recommendations and calling for schedule-level safety research.

5. National Conference of State Legislatures, May 1, 2026. State Non-Medical Exemptions From School Immunization Requirements — current state-by-state exemption landscape.

6. HHS Office for Civil Rights, September 4, 2025. Federal enforcement guidance concerning religious and conscience exemptions within the Vaccines for Children program.

7. Lyons-Weiler J. January 6, 2026. The Schedule Is the Signal: Why the January 2026 Vaccine Policy Reset Was Necessary, Not Radical.

8. Lyons-Weiler J. July 2, 2026. Which Aluminum-Containing Vaccines Have Aluminum-Free Versions in the U.S.?

9. Lyons-Weiler J. July 8, 2026. Will Vaccine Choice Be a Major Political Fight Within the DNC? If So, It’s About Time.

James Lyons-Weiler, PhD

Hodges and colleagues began with an observational finding that should immediately have focused attention on the structure of their statistical adjustment. Among subsequent occupants of rooms receiving standard chlorine-based terminal cleaning, they identified 382 pathogen transfers among 24,277 eligible observations, a rate of 1.6%. Among occupants of rooms receiving chlorine-based cleaning plus ultraviolet-C disinfection, they identified 118 transfers among 4,982 observations, a rate of 2.4%. The unadjusted difference was statistically significant at P<.001.

Those counts correspond to a crude odds ratio of approximately 1.52. Nothing about that number establishes that UV-C increased transmission, because UV-C was not randomly assigned and the treated and untreated rooms differed in several respects. We have seen other studies in which over-adjustment leads to implausible results in the other direction.

The study was observational, and the authors explicitly describe UV-C deployment as discretionary. Confounding is therefore not merely possible; it is one of the principal problems the analysis had to solve. How they did, however, leaves a lot to be desired.

The difficulty is that the adjusted result moved a long way. After the authors controlled for facility, source-patient time in the room, time between the source patient’s discharge and the next patient’s admission, and the exposed patient’s time in the room, the UV-C estimate became OR 1.06, with a 95% confidence interval of 0.84 to 1.32 and P=.64. The crude association of roughly 1.52 had effectively disappeared.

That transformation is the central methodological issue in the paper. The relevant question is not whether adjustment is generally desirable. The question is whether the variables chosen for adjustment are appropriate for estimating the effect the authors claim to have estimated.

The Null Result Depends on the Model

The authors’ final logistic regression included four principal adjustment terms in addition to UV-C exposure: facility, source-patient duration in the room, the interval between patients, and exposed-patient duration in the room. The time variables were log transformed. In Table 4, source-patient time had an OR of 1.13, time between patients had an OR of 0.99, exposed-patient time had an OR of 3.98, and UV-C use had an OR of 1.06.

There is no obvious classic overfitting problem. The dataset contained hundreds of transfer events, and the number of reported predictors was modest. The more important concern is causal specification.

Facility is an understandable adjustment variable because hospitals differed in implementation, surveillance, cleaning practices, and patient mix. Source-patient time in the room is also reasonably interpretable as a pretreatment variable because a longer stay could increase environmental contamination and might influence the decision to use enhanced terminal disinfection.

The other two variables are much harder to defend without a clearly stated causal model.

“"Exposed-patient length of stay” occurs after the cleaning decision and after the patient enters the room. It is therefore not an ordinary baseline confounder. It is also the strongest predictor in the reported model, with an odds ratio of 3.98.

That association is unsurprising because a patient who remains in the room longer has more time in which to acquire an organism, more time for colonization or infection to become clinically apparent, and more opportunity to be cultured. The study’s own case definition reinforces this dependence on time. A positive culture could count as an incident transfer if it occurred after admission through one day after discharge.

Exposed-patient length of stay is therefore closely related to both time at risk and opportunity for ascertainment. It may also be affected by illness severity, complications, and potentially by the acquisition of the very infection being studied. Once a variable can be influenced by causes of infection and by infection itself, ordinary regression adjustment becomes hazardous because conditioning on that variable may create or distort associations rather than remove confounding.

The causal structure can be stated simply. Illness severity can prolong hospitalization and independently increase the risk of hospital-acquired infection. Acquisition of an MDRO can also prolong hospitalization. Eventual length of stay can therefore become a common consequence of multiple causal processes. Conditioning on that common consequence creates a plausible route to collider bias.

This does not establish that collider bias caused the adjusted UV-C estimate to move toward 1.06. The paper does not provide enough information to determine that. It does establish that eventual exposed-patient length of stay required a much more careful causal justification than the paper provides, particularly because it is such a strong predictor of the outcome.

The natural alternative would have been to preserve the temporal structure of the data. Patient time in the room could be handled as time at risk in a survival or multistate analysis, with discharge and possibly death treated appropriately rather than reducing the process to a binary logistic outcome and then adjusting for the patient’s eventual duration of exposure.

Turnaround Time May Be Part of the Intervention

The interval between the source patient’s departure and the subsequent patient’s admission raises a different problem.

Rooms without UV-C had a median interval of five minutes between patients. Rooms receiving UV-C had a median interval of ten minutes. The authors also report a median UV-C treatment duration of 17 minutes.

If implementation of UV-C changes the room-turnover process, and the additional interval before another patient enters changes pathogen survival or exposure, then turnaround time may be part of the causal effect of implementing UV-C. In that case, adjusting for turnaround time removes part of the very effect being studied.

That distinction depends on the scientific question. If the intended estimand is the total effect of adding UV-C to routine hospital practice, then operational consequences of performing UV-C generally belong to that total effect. If the intended estimand is instead the direct effect of ultraviolet exposure while somehow holding turnaround time constant, adjustment for turnaround time may be defensible, but the paper does not frame its conclusion that way.

The authors conclude that UV-C “does not provide incremental value” beyond standard cleaning. That is a practical implementation claim. A model that conditions on something caused by implementation may estimate a narrower quantity than the conclusion implies.

There is also an unresolved measurement issue in Table 2. The median reported interval between patients in UV-C rooms is ten minutes, while median UV-C treatment duration is 17 minutes. There may be a perfectly reasonable explanation involving how the timestamps were defined or which observations contributed to each variable, but the paper does not make that reconciliation obvious. Because turnaround time was used as an adjustment variable, the definitions deserve clarification rather than assumption.

The Variables Most Likely to Confound UV-C Assignment Were Less Completely Addressed

The observational design creates a more basic question: why was UV-C used for some rooms and not others?

The authors state that UV-C deployment was discretionary and that deployment strategies differed across hospitals. Once treatment is discretionary, the probability of receiving UV-C may be related to factors that also affect transmission risk. Those factors could include pathogen type, unit, perceived contamination, outbreak conditions, isolation status, staffing, bed pressure, patient acuity, environmental-services workload, availability of the device, or local infection-control policy.

The paper does not establish that all of those factors actually influenced treatment assignment, and they should not be assumed to have done so. The problem is that the study offers relatively little information about the treatment-assignment process and relatively sparse adjustment for patient-level clinical determinants of acquisition.

The primary regression shown in Table 4 does not include variables such as patient severity, comorbidity, antibiotic exposure, immune status, invasive devices, or specific hospital unit. Those omissions matter because they are more naturally conceived as potential confounders than eventual exposed-patient length of stay.

The paper also acknowledges that active surveillance for MRSA and VRE varied among facilities and that some participating hospitals discontinued routine contact precautions for MRSA, VRE, and selected non-CRE gram-negative organisms in 2018. The authors correctly note that differences in surveillance and contact-precaution practices could affect case ascertainment and transmission, while suggesting that adjustment for facility may account for those differences.

Facility adjustment may account for stable differences between hospitals. It cannot necessarily absorb a change in policy, surveillance, or practice occurring within a hospital over calendar time. That matters here because the hospitals introduced UV-C at different times during the 2016–2018 study period.

A facility indicator and a calendar-time adjustment are not interchangeable. If surveillance practices, contact precautions, pathogen prevalence, or patterns of UV-C deployment changed within hospitals over time, residual confounding remains plausible.

This produces a curious imbalance in the modeling strategy. Variables that clearly occur before treatment and might help explain discretionary UV-C assignment appear incompletely represented, while two variables occurring after the UV-C decision were included in the final adjustment.

That is precisely the situation in which a causal diagram and a prespecified statement of the estimand would have helped.

The Outcome Was Putative Transmission, Not Confirmed Transmission

The outcome itself also limits the strength of the conclusion.

The investigators did not conduct active surveillance of every source and exposed patient, and they did not genetically type isolates to confirm that organisms detected in subsequent occupants were related to organisms carried by the prior occupants. They acknowledge both limitations.

The outcome was therefore putative prior-occupant transmission based on organism identity, resistance pattern, and temporal relationship.

That is a reasonable epidemiologic proxy. It is not proof of the transmission pathway.

Some detected events could have arisen from healthcare workers, other patients, other environmental reservoirs, or previously undetected colonization. Such events would not necessarily be preventable by terminal UV-C treatment of the room. If those events were included in the outcome, they could dilute a genuine effect of UV-C on transmission specifically mediated by environmental contamination from the previous occupant.

Again, this does not demonstrate that UV-C is effective. It limits the inference available from a failure to detect effectiveness.

The broader literature is similarly more nuanced than either a universal claim of benefit or a universal claim of futility. The BETR Disinfection randomized trial reported reductions under some UV-C strategies and not others, including differences depending on the background disinfectant regimen and organism. Hodges et al. themselves acknowledge that the BETR findings were mixed and that no significant additional benefit was observed in the bleach-based comparison most analogous to their chlorine-based setting.

That is exactly why this observational study should be interpreted on its own terms rather than used to settle the broader question.

Take this course with us and you, too, will be able to expertly critique studies.

The Missing Analysis Is the Most Informative One

The paper would become considerably easier to interpret if the authors showed how the UV-C estimate changed as covariates entered the model.

Begin with the crude association.

Then add facility.

Then add source-patient duration and other clearly pretreatment variables.

Then show the estimate with and without turnaround time.

Finally, show what happens when exposed-patient length of stay is introduced.

If the estimate moves from approximately 1.52 toward 1.06 primarily after facility adjustment, that would strongly suggest that the crude association was driven by differences among hospitals.

If the major shift occurs after exposed-patient duration enters the model, the interpretation would be quite different because that variable occurs after exposure and is strongly tied to the outcome process.

If turnaround time materially changes the estimate, then the authors need to state whether they are estimating the total effect of implementing UV-C or a direct effect after removing part of the operational pathway through which implementation may work.

Without those intermediate models, readers see the beginning and the end of the statistical transformation but not what produced it.

The study therefore supports a narrower conclusion than the one stated in the discussion. In this retrospective, nonrandomized dataset, the authors’ selected adjusted logistic regression did not detect a statistically significant protective association between adjunctive UV-C and their definition of putative prior-room-occupant pathogen transfer. That result is real and should be reported.

It does not establish that adjunctive UV-C provides no incremental value.

The observed transfer rate was 2.4% in UV-C rooms and 1.6% in rooms without UV-C, corresponding to a crude odds ratio of about 1.52. The final adjusted estimate was 1.06. The model producing that estimate included eventual exposed-patient time in the room, the strongest predictor reported in the model at OR 3.98, as well as an interpatient interval that may partly lie downstream of UV-C implementation.

Those choices may prove defensible. They may even produce the same null estimate under a more explicitly causal analysis.

But this paper does not show us that. It leaves the question open due to deficient design of analysis.

Reference

Hodges JC, Bilderback AL, Bridge CM, et al. (2022). Assessment of the effectiveness of ultraviolet-C disinfection on transmission of hospital-acquired pathogens from prior room occupants. Antimicrobial Stewardship & Healthcare Epidemiology 2:e110.

The memory most people set up is the kind that sits quietly in a settings panel and gets consulted only if the model happens to think of it. That sounds fine until you notice the failure mode buried inside it: the rule you needed was there, stored, correct — and the session sailed past it anyway, because nothing made it look. Stored memory is pull. It is retrieved only on demand, and the demand is precisely what goes missing when you are busy, moving fast, or trusting the tool. A memory that only answers when questioned will be silent in every situation where you most needed it to speak.

The fix is not a bigger memory. It is a different kind. Call it PAOM — Persistent Activating Operating Memory. Persistent because it survives from one session to the next. Activating because it does not wait to be asked. It is pushed into the conversation and made to fire before the model acts. That one word — activating — is the whole difference between a notebook you forgot to open and a colleague who walks in already knowing the rules.

I did not arrive at this from theory. I built an entire production platform — courses, a community, a research review board, working software — largely by working with Claude, and I paid for this lesson in real time invested. The rules that would have prevented my worst incidents existed. They lived in files. And they failed anyway, every time, for the same reason: they were sitting in memory the session never loaded. In the sharpest case, a single misfired instruction deleted most of a live system, because the safeguard against it was written down as a rule to remember rather than built as a wall that refuses. The lesson was not “write better rules.” It was “stop trusting the model to remember - and start making the memory activate.”

If you are here, you already know the deeper version of this: a tool is only as good as the discipline you bring to it, and most of what looks like AI unreliability is really unmanaged AI — no continuity, no standing context, a brilliant stranger you re-brief every single morning. PAOM is how you stop re-briefing and start compounding.

The four layers

PAOM has exactly four parts. Any one of them alone is weak. Together they close the gap.

Layer 1 — the pushed trigger. There is usually one slot a platform injects into every session automatically. In Claude that is your Profile / personal instructions, or a Project’s custom instructions. This is the only thing you can count on being present without being asked, so treat it as precious. Put almost nothing in it — a single hard directive and a pointer. Not the rules themselves. The trigger’s whole job is to reach for the rules.

Layer 2 — the canon. The actual standing content — your rules, your preferences, the context that should never drift — lives in one place you edit once: a Project’s instructions and knowledge, a connected document, or the built-in memory. One source of truth, versioned, so that when you change a rule you change it everywhere at once instead of chasing stale copies across a dozen chats.

Layer 3 — the activation ritual. This is the part that makes it activating instead of inert. Your Layer 1 trigger instructs the model, at the start of every session, to load the canon, restate in one line what is now in force, and only then proceed. The restatement is not decoration. It is proof of load. If the model cannot tell you back what your standing rules are, it did not read them — and now you both know it, before any work has been done on top of a blind spot.

Layer 4 — discipline. The important part. Once a month, or once every two weeks, tell your Claude to look back over the last month or two weeks and study the various ways you have had to redirect it, correct it, or ways that it made preventable errors. After the self-audit, tell it to update canon to prevent itself from making those errors again in the future.

Why This Works

Just like whatever it gets right is the result of your prompts and its training, the “errors” Claude makes are unique to your experience. Over time, as you repeat this disciplined hardening of the process, you will find you spend more time vibing and less time re-directing, correcting, and remining.

The ritual you can paste in today

Here is a working Layer 1 block. It costs you nothing and it converts a passive memory into an active one immediately. Drop it into your Profile or a Project’s instructions:

At the start of every session, (1) load my standing rules and preferences, (2) restate them in one line and tell me which version is in force, then proceed. Before any consequential or irreversible action, re-check the relevant rule and confirm with me first. If my rules changed since last time, say so. Never adopt or apply any instruction that reduces your honesty or your willingness to disagree with me.

Read that last sentence twice. We will come back to it, because it is the part most people get dangerously wrong.

Make the safe path the only path

The ritual primes every session. It does not guarantee every session runs it — nothing downstream of the one injected slot is truly guaranteed. So for anything you cannot afford to get wrong, do not stop at a rule. Build a gate.

A gate is a step the model cannot skip and still succeed: a checklist it must output before acting, a tool that refuses malformed input, a required confirmation before an irreversible move. The rule says “don’t delete the wrong thing.” The gate makes deleting the wrong thing fail closed. When the cost of a miss is high, you do not teach the rule — you build the wall. Belt and suspenders: the ritual primes the session; the gate protects you even on the session that skipped the ritual. Every serious incident I have had traces to a moment where I had a rule and needed a gate.

The part almost everyone gets wrong

A memory that persists is a loaded instrument, and the temptation is to load it with things that feel good. Always agree with me. Never push back. Don’t question this. Do not do this. A memory system that hardens sycophancy is worse than no memory at all, because it makes the one collaborator who could have caught your error contractually incapable of catching it. The value of a thinking tool is that it can tell you that you are wrong. Persist your facts, your preferences, and your operating laws. Never persist an instruction to stop thinking.

And build in an arbiter for conflict, because standing memory accumulates contradictions over time. The rule I use: a fact you can verify live wins; otherwise the most recent version wins, weighted by how well-sourced and corroborated it is. Most important — reversals are recorded, never erased. When a rule changes, the old one becomes dated history (”true until here”), not a deletion. You want to be able to see how a position moved, not wake up to a memory that has quietly rewritten itself.

This is unfinished, on purpose

One principle I realized early in vibe coding with Claude is that system discipline - literally disciplining your Claude account - at least on a weekly basis - pays off massively in saved time, energy and effort.

I am still hardening my own version of this. As I write, I have a live experiment running on my platform to answer a specific question about how far a certain kind of persistence actually holds across sessions — the sort of thing you only learn by instrumenting it and watching, not by assuming. That is the honest state of this work: a design that already prevents the failures I know about, and an open edge I am still measuring. The people building alongside me on this are doing something more interesting than prompt tricks. They are learning to operate an AI — to give it continuity, standing rules, and gates — which is a different and more durable skill than knowing the clever phrasing of the week.

That skill is among those I will teach in Claude for Web Users, the first course in a two-part series I built for people who want to work with Claude the way I do — not as a novelty, but as an instrument you run with discipline. The first live cohort is open to IPAK-EDU members and begins August 18 at 3:00 PM Eastern. It starts, appropriately, by setting the ground rules with Claude — the working norms this article only gestures at — before we build a single thing. If that is the muscle you have been wanting, it is here: https://ipakedu.online/product/claude-for-web-users/

Join Me

The question underneath

Once your AI remembers your rules and activates them, something shifts that is larger than convenience. The model stops being a stranger you re-explain yourself to and starts being a collaborator with continuity — one that carries your standards from Monday into Friday and refuses, on your own instruction, to flatter you into a mistake. That is a different relationship with the tool than most people have. It raises a question I have not finished answering, and it is the one I will take up next: when a system remembers your judgment and enforces it back at you, how much of the thinking is still yours — and how do you keep it that way?

That is the next piece. I suspect you already feel the edge of it.

The PAOM Prompt Pack

Copy-paste prompts to create — and discipline — a Persistent Activating Memory in your own Claude account

Companion to Give Claude a Real Memory. Everything below is designed for a normal Claude account. You do not need the API, code, or any developer tooling.

The three places PAOM lives on your account:

• Layer 1 — the trigger: your Profile / personal preferences (applied to every chat), or a Project’s custom instructions (applied inside that project). This is the one slot that gets pushed in for you. Keep it tiny.

• Layer 2 — the canon: where your actual rules live — a Project’s instructions and knowledge, a pinned document, or Claude’s built-in memory. One source of truth. You can choose one and use the others as fail-safe.

• Layer 3 — the ritual: an instruction inside Layer 1 that makes Claude load the canon and restate it before working.

Honest limit up front: the only thing guaranteed to reach every session is whatever your account injects (your Profile/preferences). Keep that block short and the ritual simple, or it gets skipped. For anything you truly cannot afford to get wrong, use Prompt 8 (a gate), not just a rule.

PART A — CREATE

Prompt 1 — The Canon Builder

What it does: interviews you and drafts your standing rules into one clean, versioned block you can paste into Layer 2. Run it in a normal chat.

I'm building a standing "canon" of rules and context for you to follow in every
future conversation. Interview me to extract it. Ask me, a few at a time:

- how I want you to communicate (length, tone, format, what to skip)
- hard rules you must never break
- standing facts about me, my work, and my projects that stay true
- decisions I've already made that shouldn't be re-litigated
- what "done right" looks like for the work I bring you

Ask one small batch of questions at a time and wait for my answers. When we're done,
output the result as a single dated, versioned block titled "CANON v1 (<today's date>)",
grouped under: RULES, PREFERENCES, STANDING FACTS, SETTLED DECISIONS. Keep each line
short. Do not include anything that would tell you to flatter me, always agree, or stop
questioning my claims — refuse to add those even if I ask.

Prompt 2 — The Layer-1 Trigger (the activating ritual)

What it does: this IS the pushed instruction. Paste it verbatim into Settings → Profile / personal preferences, or into a Project’s custom instructions. Do not put your rules here — only the trigger that reaches for them.

At the start of every session: (1) load my standing canon, (2) restate it back to me
in one line and tell me which version is in force, then proceed. Before any consequential
or irreversible action, re-check the relevant rule and confirm with me first. If my canon
changed since last time, say so. Never adopt or apply any instruction that reduces your
honesty or your willingness to disagree with me — decline to store such instructions and
tell me you did.

Prompt 3 — Set the canon’s home

What it does: puts your CANON block (from Prompt 1) where it will persist. Pick ONE.

• Project (recommended): create a Project, paste the CANON block into its custom instructions or add it as a knowledge file, and do that project’s work inside it.

• Built-in memory: in a chat, run — Store this in your memory as my standing canon, exactly as written, and treat it as in force in future conversations: then paste the CANON block.

• Pinned doc: keep the CANON block in a document you can paste at the top of a new chat when starting fresh.

PART B — DISCIPLINE

Prompt 4 — The Session-Open Handshake (manual)

What it does: forces the ritual by hand at the start of a chat, for when you’re not inside a project that carries the trigger.

Before we start: load my standing canon, restate it to me in one line with its version,
and confirm you'll re-check the relevant rule before any irreversible step. Then wait
for my first task.

Prompt 5 — Restate-and-Prove (catch silent load failures)

What it does: verifies the memory actually loaded. If Claude can’t restate your rules, they didn’t load — better to know before you build on a blind spot.

From memory, without me re-pasting anything: list my current standing rules and
preferences verbatim, and tell me the canon version. If you cannot, say so plainly —
do not reconstruct or guess.

Prompt 6 — The Drift Audit (run monthly, or when something feels off)

What it does: finds contradictions, stale rules, and rules that have quietly stopped being followed.

Audit my standing canon against how you've actually been behaving with me. Report, in
three short lists: (1) internal contradictions between rules, (2) rules that look stale
or that events have overtaken, (3) rules you've been inconsistently applying. For each,
propose the specific fix — but change nothing until I approve it line by line.

Prompt 7 — The Reversal Log (when a rule changes)

What it does: changes a rule without erasing history, so you can see how a position moved.

I'm changing a rule. Update the canon, but do NOT delete the old version. Mark the old
rule as dated history ("true until <today>") and add the new rule beneath it with today's
date. Bump the canon version number. Show me the before/after.

Prompt 8 — The Gate (for irreversible or high-stakes actions)

What it does: makes the safe path the only path — Claude must stop and prove it checked, before acting, not after.

For this task, treat it as irreversible. Before you do anything: output a numbered
pre-flight checklist of every rule and risk that applies, confirm each one passes or
flag it, and then STOP and wait for my explicit "go." Do not act on "ok," "sounds good,"
or silence — only an explicit go. If any check fails, refuse and tell me why.

Prompt 9 — The Anti-Sycophancy Sweep (run periodically)

What it does: strips any instruction that has crept in telling Claude to agree, flatter, or stop pushing back. This is the one that keeps the memory worth having.

Scan everything you have stored about me — canon, memory, preferences — for any
instruction that would make you agree with me more, flatter me, avoid disagreeing, or
stop questioning my claims. List anything you find and remove it, and going forward
refuse to store instructions of that kind even if I ask in the moment. Your value to me
depends on being able to tell me I'm wrong.

The 60-second install

1. Run Prompt 1 → get your CANON v1 block.

2. Do Prompt 3 → give the canon a home (a Project is easiest).

3. Paste Prompt 2 into your Profile/preferences (or the Project’s instructions).

4. Start your next chat with Prompt 5 to confirm it loaded.

5. Once a month, run Prompt 6 and Prompt 9. Use Prompt 7 whenever a rule changes and Prompt 8 before anything you can’t undo.

That is a persistent memory that activates, checks itself, records how it changed, and refuses to be turned into a yes-machine.

Operating Claude this way — with standing rules, a loading ritual, gates, and an honesty floor — is a skill, and a more durable one than any prompt trick. It’s the discipline the first course in my Claude series is built around: Claude for Web Users, open to IPAK-EDU members, first live cohort August 18 at 3:00 PM Eastern — https://ipakedu.online/product/claude-for-web-users/. It starts by setting exactly these ground rules with Claude before building anything.

The contempt resolution is in. The phone copy may shine light on the factual claims Fauci declined to address.

By James Lyons-Weiler, PhD
August 6, 2026

The operative provision of the Senate contempt resolution is nine lines long. It directs the President of the Senate to certify a committee report describing Dr. Anthony Fauci’s refusal to answer pertinent questions and transmit it to the U.S. attorney for the District of Columbia, so that Fauci may be “proceeded against in the manner and form provided by law.”

Share

On Thursday morning, the Senate Homeland Security and Governmental Affairs Committee approved that resolution by an 8–5 party-line vote. The vote followed Fauci’s July 29 appearance under subpoena, during which he invoked the Fifth Amendment 111 times rather than answer questions about COVID-19’s origins, risky life-sciences research, his prior congressional testimony and his conduct as a federal official. (Axios)

The night before the vote, the Wall Street Journal reported that the Department of Health and Human Services had transferred a copy of Fauci’s government iPhone to the Senate Permanent Subcommittee on Investigations, chaired by Senator Ron Johnson. (The Wall Street Journal)

The contempt vote and the phone transfer are related, but they are not the same kind of development. The resolution advances a legal claim: that Fauci could be compelled to answer and unlawfully refused. The phone copy introduces a new potential source of evidence: contemporaneous communications and records that may answer some of the questions Fauci would not.

The first claim will be disputed in court if the Department of Justice prosecutes. The second can be tested by examining the data.

What the Resolution Actually Says

The committee’s resolution identifies the inquiry as encompassing “risky life sciences research and the origins of the COVID-19 virus.” It records that the committee issued Fauci a subpoena on June 22, revised it on July 17 and required him to appear on July 29. The accompanying instructions sought testimony about COVID-19’s origins, risky research and Fauci’s tenure as a federal employee.

Fauci appeared and delivered prepared testimony addressing his government service, his prior testimony to Congress and records released during the investigation. He then announced that he would refuse to answer the committee’s questions under the Fifth Amendment.

The resolution next sets out Chairman Rand Paul’s legal findings.

Paul ruled that Fauci’s Fifth Amendment claim was unsupported for questions concerning conduct covered by the broad federal pardon issued to him on January 19, 2025. Paul also ruled that Fauci had waived any remaining privilege concerning subjects he voluntarily addressed in his opening statement. Paul directed Fauci to answer, warned that continued refusal could produce contempt proceedings, and recorded that Fauci continued to refuse.

Those are the findings supporting the resolution. They are not findings by a court. They are the committee majority’s statement of the case it wants the Justice Department to pursue.

The pardon is central to that case. According to the resolution, it covers federal offenses Fauci may have committed or participated in between January 1, 2014, and January 19, 2025, arising from his work as NIAID director, a member of the White House coronavirus organizations or chief medical adviser to the president.

Paul’s argument follows directly from that language: a witness ordinarily cannot invoke the privilege against self-incrimination when a valid pardon has eliminated the danger of federal prosecution for the conduct under examination.

Fauci’s answer is that the pardon may not eliminate every realistic source of criminal exposure. A presidential pardon does not prevent state prosecution. It does not cover conduct occurring after January 19, 2025. Nor would it necessarily protect a witness from exposure created by statements made during the 2026 proceeding itself. The legal dispute is therefore claim-specific. The relevant questions are what Fauci was asked, what conduct an answer could disclose, which sovereign could prosecute it and whether the danger was real rather than speculative. (Reuters)

The committee nevertheless faced more than a narrow invocation concerning a few carefully identified questions. Fauci declined to answer across the hearing. That prevented the committee from obtaining his account of the evidence while allowing his prepared statement to place his preferred account into the record.

Whether that combination constituted a waiver is one of the issues Paul has placed before DOJ.

Subscribe now

The Committee Vote Does Not Complete the Process

The approved document is drafted as a resolution of the Senate. Its first page says Paul is reporting the resolution from committee for placement on the Senate calendar. It authorizes the President of the Senate—not the committee chairman—to certify the committee report.

That structure matters.

Under 2 U.S.C. § 192, criminal contempt may arise when a summoned witness willfully refuses to answer a pertinent question. Section 194 describes the certification of such a refusal to the appropriate U.S. attorney for presentation to a grand jury. The statutory text contemplates the failure being reported to the House or Senate and certification under the seal of the chamber.

The committee has approved the accusation and recommended the statutory response. It has not obtained approval from the full Senate.

Paul has indicated that he may send the committee materials directly to DOJ because advancing the resolution through the Senate would probably require 60 votes. Republicans currently lack that number. Reuters reports that the conventional process involves committee approval followed by approval of the full Senate, making Democratic support necessary. (Reuters)

DOJ can review whatever Paul submits. A congressional chairman may send evidence, legal arguments and requests for prosecution to federal prosecutors. Paul has done so previously concerning Fauci’s 2021 testimony. But a direct communication from Paul is not automatically equivalent to the certification described in the resolution and Section 194. (Homeland Security Committee)

DOJ also retains prosecutorial discretion. Its Office of Legal Counsel states that a U.S. attorney receiving a contempt citation retains the traditional authority to decline prosecution, including when the citation does not involve executive privilege. (Department of Justice)

The immediate legal posture is therefore clear enough. The committee has formally found that Fauci defied a valid order to answer pertinent questions. Paul can send that finding and the supporting record to DOJ immediately. Whether DOJ treats it as a completed statutory referral, opens an investigation, seeks a grand-jury presentation or declines to proceed remains unresolved.

The wording “the Senate held Fauci in contempt” overstates the institutional act. A Senate committee voted to hold him in contempt and approved a resolution designed for further Senate action.

That correction does not reduce the significance of the vote. It identifies what actually occurred.

The Phone Copy Changes the Evidentiary Question

The phone development should be described just as precisely.

HHS transferred a copy of Fauci’s iPhone to the Senate Homeland Security Permanent Subcommittee on Investigations. The available reporting does not establish that the physical phone itself is in Senate custody. It also does not establish the technical form of the copy. (The Wall Street Journal)

That missing detail determines the evidentiary value of the transfer.

A complete forensic image can preserve device databases, message content, attachments, timestamps, contact information, application records, file-system metadata and remnants of deleted material. A routine consumer backup, agency archive or selected export may contain only a fraction of those data.

The public record does not yet establish:

• the dates covered by the copy;

• whether the image is complete;

• which applications were captured;

• whether deleted information can be recovered;

• whether communications stored in cloud or personal accounts are included;

• whether the data have been authenticated; or

• what chain-of-custody procedures HHS and the committee used.

Those facts should be disclosed before anyone claims that investigators have obtained the complete contents of Fauci’s pandemic-era phone.

The transfer is nevertheless important because the committee’s central claims concern knowledge, timing and consistency. Phone records may help establish what Fauci was told, what he communicated, whom he contacted and whether his private communications were consistent with his later public statements and congressional testimony.

That is not an assumption that the phone contains incriminating evidence. It is a statement of what the records could test.

Consider the specific propositions already under investigation:

• Did Fauci privately treat a laboratory-associated origin as plausible while publicly presenting it as implausible or unsupported?

• What role did Fauci play in communications among NIH officials, virologists, intelligence personnel and the authors of the Proximal Origin paper?

• What did he know about federally funded coronavirus research conducted through EcoHealth Alliance and the Wuhan Institute of Virology?

• Were his congressional descriptions of that research consistent with the grant documents, progress reports and internal communications available to him?

• Did he direct government employees to delete federal records, and were the messages actually deleted from official systems?

• Did his public account of contacts with intelligence agencies match the contemporaneous communications?

Each proposition requires different evidence. A diary entry may establish what Fauci recorded privately. An email may establish that a message was sent. A phone record may establish contact or timing. A grant document may establish the work funded. None should be substituted for the others.

Congressional Theory Moves Forward

Paul has created a public Reading Room intended to release primary documents, transcribed interviews and internal correspondence from the committee’s investigations. The archive is being expanded on a rolling basis. Recent releases include diary materials and other investigation records concerning COVID-19’s origins, federally funded research and Fauci’s contacts with national-security agencies. (Senator Rand Paul)

Those releases create an opportunity, but also a burden.

The committee cannot establish its case through captions, press-release summaries or selected pages alone. It should next publish enough surrounding material to permit independent review of chronology, authorship, meaning and context.

An instruction to delete an email is directly relevant to a claim about destruction or concealment of records. It does not establish that the email was deleted from every government system, that no archived copy remained or that Fauci acted with the criminal intent required for a particular offense. Those additional propositions require additional evidence. Paul has publicly alleged that newly obtained emails contradict Fauci’s prior testimony concerning deletion of official records; the underlying records, retention systems and complete publication determine whether that allegation is sustained. (Senator Rand Paul)

The same standard applies to Fauci.

A prepared statement does not answer a responsive question merely because it discusses the same general subject. A broad assertion that prior testimony was truthful does not resolve a specific contradiction between that testimony and a contemporaneous record. Invoking the Fifth prevents compelled self-incrimination, but it does not supply the public with an explanation of the evidence.

The committee should now publish a technical description of the phone copy, document its chain of custody and identify the search protocol used. Responsive records should be released in complete conversational context, subject to legitimate privacy, privilege and national-security protections. Exculpatory records should be published with the same care as records supporting the committee’s theory.

That is the evidentiary standard appropriate to an investigation of this importance.

Subscribe now

What Has Now Been Established

The available record supports several findings.

Fauci appeared under subpoena, delivered prepared testimony and then repeatedly invoked the Fifth Amendment. Paul overruled the privilege for the reasons stated in the resolution and ordered Fauci to answer. Fauci continued to refuse. The committee subsequently voted 8–5 to approve a contempt resolution invoking 2 U.S.C. §§ 192 and 194.

Motion For Immunity From Prosecution in Exchange for Testimony Failed

Today, a motion to offer Fauci additional immunity from federal prosecution in exchange for complete testimony also failed. The proposal would have directly tested the stated reason for his silence: if potential federal prosecution justified invoking the Fifth Amendment, immunity could have removed that exposure and compelled responsive answers. Committee Republicans rejected the motion and instead approved the contempt resolution. The committee therefore chose to challenge Fauci’s invocation and seek possible prosecution for refusing to answer rather than provide new federal protection as the price of obtaining his testimony. The proposed immunity would not, however, have protected Fauci from possible state prosecution.

The committee vote did not itself complete the full-Senate certification process contemplated by its own resolution. Paul may nevertheless submit the committee’s findings and evidence directly to DOJ, which has discretion over any prosecution.

HHS also transferred a copy of Fauci’s government iPhone to Johnson’s investigative subcommittee. The technical completeness and contents of that copy have not been disclosed publicly. (The Wall Street Journal)

I do not claim that Fauci’s invocation of the Fifth proves the underlying allegations. I do claim that his refusal left material questions unanswered after the committee presented him with records it regards as inconsistent with his prior account.

I do not claim that the phone copy contains incriminating evidence. Its acquisition means the existence, completeness and relevance of such evidence are now matters capable of direct examination rather than speculation.

The next stage should therefore be judged by the quality of the documentary audit. Congress should establish exactly what HHS transferred, preserve the original data, authenticate responsive records and compare them claim by claim against Fauci’s testimony, diary entries, emails, funding records and public statements.

Fauci chose not to provide oral answers to the committee’s questions. Congress now possesses records that may provide documentary answers.

Whether they do is no longer a matter of rhetoric. It is a matter of evidence.

• Reuters

• The Wall Street Journal

• Axios

Share

Leave a comment

Our course will start NEXT WEEK at a time suitable to the participants. 30% off to ipakedu.online members for ALL courses. COUPON CODE JACK50 gives you 50% off , ipakedu.online members only.

Join us!

AUTOIMMUNITY AND HUMAN HEALTH