The Deadline Moved. Six Things Did Not.

Europe has just given international arbitration sixteen more months. It has not given you a reprieve, and the decision that will actually govern your practice was taken while everyone was watching the calendar.

Prof. Barry Appleton, New York Law School | TechCred | 28 July 2026

Where this stands as at 28 July 2026. The Digital Omnibus on AI was adopted by the Council on 29 June 2026, following the European Parliament’s endorsement on 16 June. It defers the stand-alone high-risk obligations under Annex III of the AI Act from 2 August 2026 to 2 December 2027. Arbitration sits in Annex III. If you are reading this later, that date is the one to check first.

For the past year, arbitration practitioners who were paying attention had 2 August 2026 circled. That was the day the high-risk regime under the EU AI Act was to apply, and with it Article 86, which entitles a person affected by a decision taken on the basis of a high-risk system to a clear explanation of the role that system played.

That date has moved to 2 December 2027.

Over the next fortnight you will hear a great deal of relief about this. Most of it will be misplaced, and acting on it will cost some practitioners a considerable amount of avoidable difficulty.

What moved, and what the move did not touch

The Omnibus deferred the compliance date. It did not defer the classification date.

On 19 May 2026 the European Commission published its draft guidelines on the classification of high-risk AI systems. The consultation closed on 23 July 2026, five days before this post. The guidelines will be finalized long before December 2027, and they will settle the two questions that matter more than the deadline ever did: which uses of these systems in arbitration are caught, and who carries the obligation when they are.

Whether arbitration is in at all is no longer seriously contestable. Paragraph 426 of the draft defines alternative dispute resolution as “adjudicating a dispute out of court with the assistance of an impartial ADR body,” and gives arbitration as an example. Recital 61 confines the limb to mechanisms producing legal effects for the parties. An arbitral award produces legal effects. Arbitration is in, and the Commission is right that it is in.

Everything interesting is downstream of that.

Six things that did not move

1. The personal mandate. An arbitrator may take help. An arbitrator may not transfer the decision. The Silicon Valley Arbitration and Mediation Center guidelines say the mandate cannot be delegated. The Chartered Institute of Arbitrators guideline says the same. None of this depends on the AI Act, and none of it has a commencement date in 2027.

2. ARIHQ c. Santé Québec, 2026 QCCS 1360. In April 2026 the Superior Court of Quebec annulled an arbitral award because artificial intelligence had generated authorities that did not exist and the award was built upon them. The vice was the delegation of the decision, not the use of a tool. Binding in Quebec today and persuasive everywhere. The Court was careful to limit itself: it said expressly that not every award citing erroneous authority, or using these tools in drafting, should meet the same fate. What made this breach fatal was that the fabricated authorities were the only legal support the award had.

3. Lnu v. Blanche (9th Cir. 3 June 2026). While Brussels was deferring, a United States court of appeals was suspending. Two attorneys who filed briefs citing decisions that never existed were sanctioned, suspended from practice before the court for six months, and their firm placed under a two-year obligation to certify in every filing whether generative tools were used, which tool, and that the signing lawyer personally verified every citation and quotation.

4. ABA Formal Opinion 512. Live since 29 July 2024. It requires American lawyers to apply an appropriate degree of independent verification to generative output, and requires supervising lawyers to train those they supervise in the capabilities and limitations of the tools. Most American arbitrators are American lawyers.

5. Article 50 transparency and Article 4 literacy. The Omnibus left the Article 50 transparency obligations on their original schedule, and the AI literacy obligation in Article 4 has applied since February 2025. Add to that data protection law, breach notification, your own professional conduct rules, and the institutional notes on AI use. A deferral of one chapter is not a holiday from the rest.

6. Your file. Which brings us to the trap.

The trap: arbitrations that straddle the deferral

This is the point that the relief is going to cost people.

An arbitration commenced this autumn will, on any ordinary timetable, still be running in December 2027. Many will be. A case that begins under a deferred regime and ends under a live one is not protected by the deferral, and the obligations that arrive in December 2027 will arrive in respect of a proceeding whose record was built in 2026.

You cannot construct a contemporaneous record retrospectively. That is what “contemporaneous” means. If an explanation is ever demanded of what a tool did and what you did, the answer is easy to give if you noted it at the time and impossible to reconstruct afterward.

The deferral gives you sixteen months to build good habits on cases where the stakes of getting it wrong are lower. It does not give you sixteen months of files you will never have to explain.

The real question: which uses, and whose obligation?

Paragraph 428 of the draft guidelines observes, correctly, that the Act does not define which entities constitute “ADR bodies.” Paragraph 429 then includes commercial arbitration institutions within that notion.

On 15 July 2026 a group of ten leading arbitral institutions with strong links to the Union filed a joint letter in the consultation, coordinated by the LCIA under the leadership of Professor Maxi Scherer. Their point is narrow and, in my view, plainly right: arbitral institutions administer proceedings. They do not decide the merits. Treating the institution as the relevant ADR body for its administrative services attaches the obligations to the wrong actor.

Professor Scherer has been telling this community to pay attention to the AI Act since 2024, when almost nobody was, and she pressed practitioners to file in the consultation rather than wait. The profession is in her debt on this subject. I supported their request in my own filing and said the guidelines should grant it.

What I filed, and why

I lodged comments with the European AI Office on 17 July 2026, in an individual academic capacity. Readers are entitled to know what I argued so they can discount for it.

I wrote in support of the draft guidelines, and to offer refinements. I did not ask for arbitration to be excluded. The credible course for arbitration is not to seek a place outside a classification that plainly fits. It is to help the Commission draw the line well.

Classify by function, not by label. The hard question is not whether arbitration is covered. It is which uses are, and the answer is functional. A system that drafts the reasoning, applies the law to the facts, or identifies authorities and proposes how to apply them operates in the high-risk zone. A system that extracts metadata, searches, proofreads or translates does not. That distinction is already in the Act: paragraph 70 of the draft guidelines makes clear that human involvement does not change what a system is intended to do, and Article 6(3) filters out systems performing a narrow procedural task or merely improving a completed human activity. A general-purpose assistant does not become high-risk because an arbitrator could open it. It becomes high-risk when it is put to, and relied upon in, the adjudicative function.

A United States federal appellate court reached the same conclusion about its own rules six weeks before the consultation closed. In Lnu v. Blanche, the Ninth Circuit explained that its procedural and ethical rules “are not violated at the point of research and drafting, but at the point of signing and filing,” that it does not matter whether a false citation came from an artificial intelligence tool “or from his own natural intelligence,” and that the rules apply “however legal papers are prepared, and however legal technology develops.” (Professor Eugene Volokh’s summary is the quickest way in.) Rules written to the function survive the technology. Rules written to the tool do not, and every list of tool categories drafted in 2026 will be obsolete before the obligations apply in December 2027.

The deployer is the arbitrator. The person who uses the system in the adjudicative function is the arbitrator, and the arbitrator therefore carries the deployer obligations. Where an institution provides administrative and organizational services, case management, scheduling, the logistics of appointment, it is not exercising the adjudicative function and should not carry deployer duties for those services.

One case needs care, and it is the constructive one. An institution can do more than administer. It can build and operate a system that itself performs the adjudicative function. The AAA-ICDR AI Arbitrator, launched in late 2025 outside the Union, drafts awards for a trained human to review before signature.

That is a careful design and the human-in-the-loop architecture is the right one. My point in the filing was not about that institution but about the drafting. Where an institution builds such a system it is the provider, and the human who adopts the output is the deployer. Each carries its own duties along the value chain, which is exactly how the Act is built to work. A classification keyed to the label of the actor rather than to the function performed would sort that case wrongly, and would do so in the one situation where the Act most needs to operate cleanly. Institutions that innovate with care should be recognized for it, and only a function-based line can recognize them.

Territorial scope has a gap, and I proposed filling it. The Act reaches deployers established in the Union, and deployers outside it where the output is used in the Union. For arbitration the guidelines do not say which connecting factor governs. My proposed answer: the relevant place for deployer status is the arbitrator’s establishment or habitual residence, not the seat, because the seat is a legal construct and not the place where a person uses a tool. The Commission should resolve that rather than leave it to be litigated.

Counsel’s tools are outside point 8(a). The provision reaches artificial intelligence used in the adjudicative function. It does not reach the tools counsel and parties use to present their case. That use is governed by professional responsibility rules and by the courts, as Lnu v. Blanche demonstrates. Where it bears on the fairness of the proceeding, the answer is the tribunal’s own duty of oversight, not treating counsel’s software as high-risk.

And nominal human involvement should not defeat the classification. Genuine judgment is required, not a signature on a machine’s output. ARIHQ is the illustration: the vice was the delegation of the decision. A clear functional line protects the enforceability of awards. It does not threaten it.

What this means for your next case

Article 3(4) defines a deployer as a natural or legal person using an AI system under its authority, other than in a personal, non-professional capacity. As Professor Scherer put it in her 11th EFILA Annual Lecture at Queen Mary in November 2025, an arbitrator using these tools professionally is a deployer of them.

Then notice what Article 2 does not contain. There is no criterion anywhere in it keyed to the seat.

Take a Delaware corporation arbitrating against a Texas corporation, New York seat, New York law, American counsel on both sides. Now add one ordinary fact: the claimant’s party-appointed co-arbitrator lives in Dublin. Nothing else about the case touches Europe. That is enough to put a deployer on your tribunal.

And a three-member tribunal deliberates jointly. Whether one member’s obligation reaches the tribunal’s collective process, or only her own contribution, is genuinely unsettled and will be argued. An arbitrator who has to argue it has already lost the year.

The tribunal secretary is a gap I did not raise in my filing, and on reflection I should have. The secretary is very often the person actually operating the tools. Some of that work, transcription, chronologies, formatting, is plainly outside the adjudicative function. Some of it, particularly research that identifies authorities and proposes how they apply, sits much closer to the line. The secretary works under the tribunal’s authority, which is the language of the deployer definition. If your secretary’s use is not recorded, the tribunal cannot give an account of it later even if it wants to, because nobody knows what was done.

What a careful tribunal does now

Four practices. Each is right on every reading of every open question above, which is the point of choosing them.

Run a connecting-factor audit at the first procedural conference. Record the establishment or habitual residence of each arbitrator and the secretary, and whether any party, witness or affected person is located in the Union. Five minutes. Almost nobody does it. You cannot apply a connecting factor you have never asked about.

Draft Procedural Order No. 1 as a tribunal undertaking, not a party permission. A permission invites the argument that what was granted could have been withheld, and therefore that what was withheld could have been granted. If the obligation turns out to be non-derogable, drafting it as a party permission concedes the case in your own document.

Bring the secretary inside the discipline, in writing, with a contemporaneous note of any tool used and for what.

Sort your own uses into the two categories now. Assistance that does not enter the reasoning, and use that supplies or shapes it. Once you have made that sort for your own practice, tiered disclosure becomes straightforward and the classification debate becomes somebody else’s problem.

The competency point, which is why this matters to you

Anyone can read a deadline off a calendar. What distinguishes a technologically competent arbitrator in 2026 is harder: the ability to hold an unsettled regulatory position and still run a procedure that is defensible whichever way it settles.

There is a second half to it, and the consultation is the illustration. Competence in a field whose rules are still being written is not only the capacity to comply. It is the capacity to participate. The rules governing how arbitrators use these systems are being drafted this year by people who will decide the question with or without us. Reading draft guidelines closely enough to find the gap at paragraph 429, and then saying so on the record, is the same skill as reading a procedural order closely enough to find the gap that will cost your client. The TechCred faculty is in that conversation because that is where the subject currently is, and I have told the Commission I remain at its disposal.

Several things remain genuinely open. Whether Article 86 travels with the Annex III deferral or applies on its own footing, given that it sits in a different chapter of the Regulation, is arguable. Whether one arbitrator’s obligation reaches a joint deliberation is arguable. Whether an award enforced in Frankfurt is “output used in the Union” for a New York arbitrator is arguable, and I have asked the Commission to settle it rather than leave it to litigation.

None of that changes what you do on Monday. Every path through every one of those questions ends at the same four practices, which is how you know they are the right four.

The institutions have made their submission and made it well. What happens to a plural tribunal when only one of its members is a deployer, and where the secretary sits in all of this, are questions the consultation has not yet reached. I will have more to say about them shortly, and in a venue better suited to the argument.

Barry Appleton, FCIArb, is Co-Director and Distinguished Senior Fellow of the Center for International Law at New York Law School, Faculty Director of the ABA International Law Section TechCred Program, and a Scholar and Fellow of the Balsillie School of International Affairs. He filed comments with the European AI Office on 17 July 2026 in the consultation on the draft guidelines for the classification of high-risk AI systems, in an individual academic capacity. Provisions cited are verified against the Official Journal text of Regulation (EU) 2024/1689. This post draws on a forthcoming article, “The Law Applicable to the Arbitral Regime.” For the institutional perspective, see the LCIA’s account of the cross-institutional initiative.

REGISTER NOW FOR TECHCRED 2026

REGISTER NOW

The Ninth Circuit becomes the first federal appellate court to suspend lawyers for filing AI-fabricated authorities, and it fixes accountability at the moment counsel signs and files.

On June 3, 2026, the United States Court of Appeals for the Ninth Circuit drew the line that two years of trial-court orders had been circling. In Lnu v. Blanche, the court suspended two lawyers from practice before it, fined them, and, as a published precedent, explained at length exactly when a lawyer becomes answerable for a fabricated citation. The answer is not subtle. The duty attaches the moment the lawyer signs and files.

For two years, the profession could treat the hallucination problem as a scatter of mostly trial-court embarrassments. District judges struck briefs and imposed fines, but those orders bound no one beyond the courtroom that issued them. The appellate courts had begun to stir. The Second Circuit referred counsel to its grievance panel in Park v. Kim, and in January the New York Appellate Division imposed sanctions in Deutsche Bank National Trust Co. v. LeTennier, the first New York appellate decision to discipline fabricated AI authorities, where counsel filed twenty-three fabricated cases across five filings and kept filing them after being put on notice. Lnu raises the altitude again. It is the first federal court of appeals to suspend lawyers from practice for the conduct, and it does so in a published opinion that binds the nine western states and the Pacific territories and persuades elsewhere. The line that runs from Mata v. Avianca, Inc. through Park v. Kim and LeTennier now carries a suspension. The two appellate decisions converge on one aggravating fact: counsel who kept standing behind fabricated authority after the court put them on notice.

What happened

The underlying matter was an immigration appeal. The petitioners, identified in the caption by the docket convention “Lnu” for last name unknown, sought review of a Board of Immigration Appeals order. Their counsel, Mike Singh Sethi and William Rounds of Orange County, filed an opening brief that cited cases that do not exist.

Two of the citations were pure inventions. The brief relied on “Eduardo v. Garland” and “Lay v. Holder”, neither of which has ever existed. Two further citations pointed to real decisions, Kamalthas v. INS and Avendano-Hernandez v. Lynch, but attached to them were quotations that those opinions never contained.

Then came the conduct that decided the sanction. When the fabrications surfaced, Sethi filed a motion describing the two nonexistent cases as “typographical errors” and asked to swap in real decisions with similar names. The replacement cases did not support the propositions for which they had been cited. Sethi did not appear at oral argument. Rounds did, and when the panel asked whether generative AI had produced the errors, he denied it three times before conceding, minutes later and only under continued questioning, that it was possible. He then revealed that the briefs had been written by an unlicensed law-school graduate whose citations no admitted attorney had read.

The court imposed a six-month suspension from practice before it, a personal sanction of $2,500 on each attorney, and an order requiring both to send the decision to their clients, to opposing counsel, and to the presiding judge in every affected matter. It went further. For two years, every filing from the firm must certify, under penalty of perjury, whether AI was used and that a responsible attorney has personally verified each citation.

What the court actually held

Read the Lnu v. Blanche opinion for what it refuses to say before reading it for what it commands. The court did not hold that lawyers may not use AI. It described such use as “not inherently unethical or irresponsible.” Sethi and Rounds were not disciplined for using a machine. They were disciplined for what they did once the machine failed them.

Four holdings carry the decision.

• The violation attaches at signing and filing. The court located the wrong not at the moment the tool drafts but at the moment the lawyer puts a name to the document and submits it. The signature is the representation. Everything that follows from a false citation flows from that act, regardless of who or what produced the words.

• The duty of candor is continuing, and a quiet fix does not satisfy it. Substituting a real case for a hallucinated one does not cure the original false statement that the fabricated case existed. A lawyer who discovers a hallucination must disclose its nature and its source. Recasting a fabrication as a clerical slip is conduct unbecoming a member of the court’s bar.

• AI errors and human errors are treated identically. A lawyer must read everything cited in a filing, whether a machine drafted it or not. The court pointed to a 2025 study finding that the leading commercial legal research tools still hallucinated answers between 17 and 33 percent of the time, and it described the resulting mistakes as the kind a first-semester law student might make but a licensed advocate may not.

• The duty is personal and cannot be delegated. Responsibility cannot be pushed down onto unlicensed staff, nor can it be pushed sideways onto the tool. Asking an AI system to check its own work is not verification. The obligation runs to the lawyer whose name appears on the brief.

Why this reaches arbitration

Lnu is a litigation decision. It governs advocates in the federal courts of the western United States, and it persuades elsewhere. It is not an arbitration case. Read it beside ARIHQ anyway.

In April, in ARIHQ c. Santé Québec, the Quebec Superior Court annulled an arbitral award because the arbitrator used AI to generate reasoning that rested on hallucinated authorities. That decision placed personal, non-delegable accountability on the arbitrator through the personal-mandate doctrine. The authority to decide belongs to the person the parties chose, and it cannot be handed to a machine.

Lnu v. Blanche places the same kind of accountability on counsel through the signature. The arbitrator who adopts a fabricated authority in an award and the advocate who files one in a memorial are each personally answerable. Neither can shelter behind the tool. Neither can shelter behind a junior. The two decisions sit on opposite sides of the same tribunal and state the same principle.

That principle travels with the person, not the forum. The duties that bound Sethi and Rounds before the Ninth Circuit are the duties of competence, diligence, and candor that bind every advocate who appears before an arbitral tribunal. Many of the arbitrators reading this also sit as counsel. The professional obligation does not switch off when the proceeding moves from a courtroom to a hearing room.

What Lnu tells every advocate and arbitrator

• Read what you cite. The duty of verification is absolute and indifferent to who drafted the text. If your name is on it, you have read and confirmed every authority in it.

• Disclose fast and disclose fully. A hallucination caught early and disclosed plainly is a correctable error. A hallucination concealed, renamed, or denied becomes a candor violation, and the candor violation is what draws the suspension.

• Do not delegate the checking. Not to a paralegal, not to an associate without a license, and least of all to the model that produced the problem.

• Use the tool, own the output. Neither Lnu nor ARIHQ bans AI. Both hold that the human who signs or who decides carries the full weight of what the tool produces.

The TechCred Program

The American Bar Association’s TechCred program certifies digital and AI competency for international dispute resolution professionals. It is the first and only program of its kind worldwide, delivered on behalf of the ABA International Law Section. Since the first class graduated in 2025, the program has taught a single discipline above all others: a competent practitioner must know when a tool is assisting judgment and when it is replacing it, must be able to detect a hallucination, and must be able to verify and disclose AI use to the parties and the tribunal.

Lnu v. Blanche confirms that discipline from the bench. We are folding it, along with every other key development in this area, directly into the next certification cohort. The next TechCred three-day international arbitration competency program runs live and online on October 22, 23, and 29, 2026.

Registration is open: https://tinyurl.com/Techcred2026

REGISTER NOW

A Final Word

The hallucination story used to be the domain of the careless few in the trial courts. After June 3, 2026, that telling no longer holds. The standard is published, binding, and precise about where responsibility lies. Not when the model drafts. When the lawyer signs.

References

Lnu v. Blanche, 177 F.4th 1014 (9th Cir. 2026) (No. 24-4790) (Paez, Bea & Forrest, JJ.), 2026 WL 1587554. Opinion: https://cdn.ca9.uscourts.gov/datastore/opinions/2026/06/03/24-4790.pdf

Association des ressources intermédiaires d’hébergement du Québec (ARIHQ) c.Santé Québec, 2026 QCCS 1360 (Que. Sup. Ct. Apr. 22, 2026) (Sheehan, J.C.S.). CanLII: https://canlii.ca/t/kkjtm

Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023) (Castel, J.), 2023 WL 4114965. Opinion: https://law.justia.com/cases/federal/district-courts/new-york/nysdce/1:2022cv01461/575368/54/

Park v. Kim, 91 F.4th 610 (2d Cir. 2024), 2024 WL 332478. Opinion: https://caselaw.findlaw.com/court/us-2nd-circuit/115760381.html

Deutsche Bank National Trust Co. v. LeTennier, 2026 NY Slip Op 00040, 2026 WL 53120 (N.Y. App. Div. 3d Dep’t Jan. 8, 2026) (Fisher, J.). Opinion: https://www.nycourts.gov/reporter/3dseries/2026/2026_00040.htm

Varun Magesh et al., Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools, 22 J. Empirical Legal Stud. 216 (2025)

Damien Charlotin, AI Hallucination Cases, https://www.damiencharlotin.com/hallucinations/

ABA TechCred Program, registration: https://tinyurl.com/Techcred2026

This is a TechCred essay drawing on a forthcoming SSRN working paper, After ARIHQ: The Non-Delegable Arbitration Mandate in the Age of AI. It is the second piece in the TechCred series on the ARIHQ decision.

When the Quebec Superior Court annulled the arbitral award in ARIHQ v. Santé Québec on April 22, 2026, much of the international arbitration community treated the decision as new law. It is not. The doctrine that decided ARIHQ — that an arbitrator cannot delegate the substantive reasoning of an award to a third party, whether human, institutional, or algorithmic — has been the rule since 1989. ARIHQ applied an old rule to a new delegate.

That distinction matters. If ARIHQ announced a new doctrine, arbitrators would be entitled to wait for further case law before adjusting their practice. If ARIHQ applied an existing doctrine to AI, arbitrators are already bound by it everywhere the doctrine runs — which is everywhere arbitration runs.

This piece traces the lineage. The forthcoming SSRN working paper, After ARIHQ: The Non-Delegable Arbitration Mandate in the Age of AI, develops the argument in full and situates the doctrine in a wider three-layer framework of arbitral governance (lex arbitri, lex data, lex AI). What follows is the doctrinal spine.

The Italian Origin: Sacheri v. Robotti (1989)

The earliest articulation of the personal-mandate doctrine in modern arbitration came not from a common-law court but from the Italian Court of Cassation. In Sacheri v. Robotti, decided in 1989, the Court held that an arbitrator who relied on a non-arbitrator expert to perform the substantive analysis on which the award rested had improperly delegated the adjudicative function. The remedy was annulment.

The reasoning has aged remarkably well. The Court did not say that arbitrators may never consult outside expertise. It said that when the outside party performs the reasoning that produces the result, the arbitrator has substituted a different decision-maker than the one to whom the parties consented. Party autonomy is the engine of arbitration. What the parties bargain for is the personal judgment of the arbitrator they selected. When that judgment is supplied by another, the bargain has not been honoured.

Sacheri established what comparative arbitration scholars would later call the rule of intuitus personae — the arbitrator’s mandate is personal to the arbitrator. It is found in Gary Born’s International Commercial Arbitration, in Gaillard and Savage’s Fouchard Gaillard Goldman, in Redfern and Hunter, and in every major treatise written in the last thirty years. The principle is not contested. The only contest is over its application.

The English Codification: P v. Q (2017)

Twenty-eight years after Sacheri, the question reached the English Commercial Court in a different form. Tribunal secretaries had become a normal feature of LCIA, ICC, and ad-hoc arbitrations. Some parties suspected that the line between administrative assistance and substantive delegation had been crossed. In P v. Q [2017] EWHC 194 (Comm), Mr Justice Popplewell took up the question directly.

Popplewell J. acknowledged what he called “considerable and understandable anxiety in the international arbitration community” about tribunal secretaries becoming “fourth arbitrators.” His ruling drew the line where Sacheri had drawn it. A secretary may perform research, prepare summaries, manage logistics, and even prepare drafts for the tribunal’s consideration. What a secretary may not do is “express a view on the substantive merits of an issue or application.” The reasoning of the award must be the tribunal’s reasoning.

The English judgment matters because it confirmed that the personal-mandate doctrine survives translation across legal traditions. Sacheri had been an Italian civil-law decision construing Italian arbitration law. P v. Q applied the same logic under section 24 of the English Arbitration Act 1996. The convergence is not coincidence. It reflects the deep structural feature of arbitration: the arbitrator’s authority is a personal delegation from the parties, and personal delegations do not sub-delegate by their own force.

Institutional codification followed. The ICC’s Note to Parties and Arbitral Tribunals on the Conduct of the Arbitration at paragraphs 184 to 194 limits the role of administrative secretaries. The Young ICCA Guide on Arbitral Secretaries (2014) makes the same distinctions in practitioner-facing form. France’s Code de procédure civile Article 1450 goes further still: “La mission d’arbitre ne peut être exercée que par une personne physique.” The arbitrator’s mandate may be exercised only by a natural person. Germany’s ZPO § 1035(5), Scotland’s Arbitration Act 2010 Schedule 1 Rule 3, and the case law of Canada, Australia, and the Netherlands all converge on the same point.

By the time ARIHQ arrived in Quebec, the doctrine had a forty-year lineage and a consolidated foundation in the leading treatises and statutes.

The Quebec Application: ARIHQ (2026)

ARIHQ v. Santé Québec did not invent the personal-mandate doctrine. It applied it to a new delegate. The arbitrator in ARIHQ had delegated the substantive reasoning of the award to a generative AI system. The reasoning rested on five citations. All five were fabricated by the AI. The award was annulled.

The four doctrinal pillars Justice Sheehan invoked — personal mandate, reasoned award, the secret of deliberation, and delegatus non potest delegare — were not Quebec inventions. They were the framework that Sacheri, P v. Q, the ICC Note, the Young ICCA Guide, and Article 1450 had together built. The first piece in this TechCred series traced the four doctrines in detail. What that piece left for this one is the genealogy.

What changed with ARIHQ is the delegate, not the rule. Tribunal secretaries, paralegal researchers, librarians, drafting assistants, and language interpreters have accompanied arbitrators for decades. None of them ever became a fourth arbitrator, because the doctrine drew the line at substantive delegation. Generative AI is a new tool with a different risk profile — its outputs are persuasive even when wrong, its hallucinations are difficult to detect, and its reasoning style mimics the reasoning style of arbitral awards. Those properties make verification more important, not less. They do not change the doctrine. They change the discipline required to apply it.

Why the Lineage Matters for Arbitrators Now

If the personal-mandate doctrine were new, arbitrators could reasonably wait for institutional guidance and further case law before adapting their practice. Because the doctrine is forty years old, the obligation is already in force. ARIHQ did not impose a new duty. It made visible a duty that has always been there.

Three operational consequences follow:

First, an arbitrator who uses a generative AI tool to draft any portion of an award must independently verify every statement of fact, every citation, and every line of reasoning before that draft becomes part of the award. The verification cannot be delegated to the AI itself. Asking a model to check its own work is not verification — it is recursion within the same delegate. The doctrine requires verification by the arbitrator personally.

Second, the arbitrator’s reasoning must be the arbitrator’s reasoning. Reasoning generated by a machine and adopted without independent judgment is not the arbitrator’s. The award may carry the arbitrator’s signature, but the personal-mandate doctrine looks at the substance of the deliberation, not the format of the signature. Sacheri and P v. Q are both clear on this. ARIHQ applied the principle to AI without difficulty because the principle was already in place.

Third, disclosure is the operational corollary of the doctrine. Where AI use materially affects the integrity of the decision-making process, the parties must be told. They cannot exercise informed consent — to the arbitrator they selected, to the procedures they agreed to, and to the award they receive — unless they know how the reasoning was produced. The disclosure obligation is not a separate rule grafted onto arbitration. It is what the personal-mandate doctrine has always required, surfaced by the new delegate’s distinctive risk profile.

The Institutional Framework Catches Up

In the year before ARIHQ, the major arbitration institutions issued AI guidance that, in retrospect, anticipated the doctrinal application Quebec would make:

• The SVAMC Guidelines on the Use of AI in Arbitration (April 30, 2024), Guideline 6: “Arbitral decision-making cannot be delegated to AI.”

• The CIArb Guideline on the Use of AI in Arbitration (March 2025, updated September 2025): “arbitral functions should not be delegated to AI.”

• The VIAC Note on the Use of AI in Arbitration Proceedings (April 2025): AI may not substitute for the arbitrator’s reasoning.

• The SCC Guide to the Use of AI (October 16, 2024): AI use raises questions about the arbitrator’s independent judgment.

• The CIETAC Guidelines on Generative AI in Arbitration (effective July 18, 2025): first AI-specific rules from a major Asia-Pacific institution; tribunals must not delegate the adjudicative function.

• The AAA-ICDR Guidance on Arbitrators’ Use of AI Tools (January 2026): disclosure required where AI materially affects the process or reasoning.

These instruments did not announce a new prohibition. They translated Sacheri and P v. Q into operational language for the AI use case. ARIHQ then translated the same principle into a judicial holding.

What This Means for the TechCred Practitioner

The personal-mandate doctrine has three implications that every TechCred-trained arbitrator should be operating against today.

The first is genealogical confidence. When a party suggests in a procedural conference that the arbitrator should be free to use AI as the arbitrator sees fit because no court has yet ruled definitively on AI delegation, the answer is that Sacheri, P v. Q, the ICC Note, and Article 1450 have all ruled, and ARIHQ has confirmed that the rule applies to AI. The doctrinal foundation does not need to be invented.

The second is disclosure-readiness. A procedural-order template that includes an AI disclosure clause, a verification protocol, and a record-of-tools provision is not a defensive document. It is a transparent application of the personal-mandate doctrine that Sacheri established. Practitioners drafting procedural orders for tribunals they sit on or appear before should treat AI disclosure language as part of the standard set, not as an optional or experimental add-on.

The third is institutional alignment. The institutional guidance has moved faster than the case law because the institutions saw the doctrinal point earlier. The SVAMC and CIArb Guidelines, the VIAC and SCC Notes, the CIETAC Guidelines, and the AAA-ICDR Guidance are not soft-law speculation. They are the operational form of an old hard-law principle. Practitioners working under those institutional rules are working with the doctrine, not around it.

Looking Ahead

The forthcoming SSRN working paper, After ARIHQ: The Non-Delegable Arbitration Mandate in the Age of AI, develops the doctrine at length and traces its consequences for enforcement under the New York Convention and the Federal Arbitration Act. It situates the personal-mandate doctrine within a wider tripartite framework of arbitral governance: lex arbitri governs procedural validity, lex data governs informational sovereignty, and lex AI governs the digital tribunal. The companion working papers on each layer will be available on SSRN in parallel.

For arbitrators reading this in the present moment, the operational message is short. The doctrine you are accountable to is forty years old. The application to AI is new. The discipline of disclosure, verification, and human judgment is what the doctrine has always required. ARIHQ did not change the rule. It made the rule unmissable.

Where TechCred Comes In

Bridging old doctrine to new practice is the curricular mission of the ABA TechCred Program, the first internationally accredited certification in AI, digital innovation, and technology dispute resolution. Sponsored by the ABA International Law Section and credentialed by New York Law School, TechCred trains arbitrators to operate the personal-mandate doctrine under the conditions that generative AI has now created — verification protocols, AI disclosure clauses, procedural-order templates, hallucination-detection workflows, and the data-sovereignty implications of using cloud-based AI tools in cross-border proceedings.

The Fall 2026 program is now open for registration.

REGISTER NOW

Further Reading

Primary sources

• ARIHQ v. Santé Québec, 2026 QCCS 1360 (Que. Sup. Ct. Apr. 22, 2026) (Sheehan J.C.S.) — canlii.ca/t/kkjtm

• P v. Q [2017] EWHC 194 (Comm) (Popplewell J.) — bailii.org/ew/cases/EWHC/Comm/2017/194.html

• Sacheri v. Robotti, Italian Court of Cassation (1989) — discussed in Born, International Commercial Arbitration (3d ed. 2021)

• French Code de procédure civile, Article 1450

• German ZPO § 1035(5)

• Scotland Arbitration Act 2010, Schedule 1, Rule 3

Institutional guidance

• SVAMC AI Guidelines (April 2024)

• CIArb Guideline on the Use of AI in Arbitration (March 2025, updated September 2025)

• VIAC Note on the Use of AI in Arbitration Proceedings (April 2025)

• SCC Guide to the Use of AI (October 2024)

• AAA-ICDR Guidance on Arbitrators’ Use of AI Tools (January 2026)

Empirical

• Damien Charlotin, AI Hallucination Cases Database (1,455 entries as of May 17, 2026)

Forthcoming

• Barry Appleton, After ARIHQ: The Non-Delegable Arbitration Mandate in the Age of AI (SSRN, May 2026)

• Companion working papers: The Lex AI, Lex Data, Arbitration Hacked, AI and International Arbitration: A Comprehensive Listing

Earlier in this series

• ARIHQ v. Santé Québec: The Case That Changed Arbitrator AI Liability

Prof. Barry Appleton, FCIArb, is the Faculty Director of the ABA TechCred Program, Co-Director and Distinguished Senior Fellow of the Center for International Law at New York Law School, and a Fellow of the Balsillie School of International Affairs. He is a dual-licensed (Canadian and American) lawyer with three decades of experience in North American trade, commerce, and investment law and a continuing scholarly focus on the intersection of AI governance and international arbitration.

Prof. Barry Appleton, FCIArb TechCred
Faculty Director, ABA TechCred Program, Co-Director and Distinguished Senior Fellow, New York Law School

This is the third piece in an ongoing series.

On June 24, 2019, a three-member arbitral tribunal in Tennant Energy LLC v. Government of Canada, dismissed the applicability of European data protection law to its own proceedings in a single paragraph. The reasoning was terse: the European Union was not a party to NAFTA, so the GDPR did not, “presumptively, come within the material scope” of the arbitration.1

That paragraph is now the textbook example of what happens when a tribunal treats data protection law as if it were optional. The arbitration was Toronto-seated. The presiding arbitrator was established in Singapore. One of the co-arbitrators was a United Kingdom-based practitioner who had publicly acknowledged acting as a data controller under the GDPR in his own data privacy notice.2 The Permanent Court of Arbitration secretariat processed case materials through European-based infrastructure.3 California-resident counsel and California-resident expert witnesses participated on the investor side, with their personal data and the personal data of California-resident witnesses moving through the proceeding throughout. The Government of Canada was the respondent. The proceeding processed personal data of witnesses, experts, and counsel located across multiple jurisdictions.

The tribunal addressed all of that with five words: “does not, presumptively, come within.”4 That answer was not merely insufficient. It was wrong. The author was lead counsel for the investor in the proceeding and is now publishing a working paper that walks through, regime by regime, why.5

Seven years later, on April 22, 2026, the Quebec Superior Court annulled an arbitral award in ARIHQ v. Santé Québec6 on grounds that included the arbitrator’s transfer of party data outside Quebec and Canada in violation of Quebec and Canadian data privacy law. Tennant and ARIHQ are now the two ends of the same arc. Tennant shows what happens when a tribunal declines to identify the applicable data protection regimes. ARIHQ shows what happens in court when those obligations are then breached. The intervening seven years have turned the lex data from an unaddressed analytical gap into an enforcement-determinative requirement.

What the Tribunal Actually Missed

The Tennant tribunal did not miss one data protection regime. It missed six. Each one attached to the proceeding through an independent jurisdictional trigger that no tribunal ruling could displace.

The GDPR. Article 3 of the General Data Protection Regulation establishes extraterritorial reach in three ways. It applies whenever personal data is processed in the Union, whenever a controller or processor is established in the Union, and whenever the targeting criteria are met. The UK-based co-arbitrator processed personal data on a personal device located in the United Kingdom (then still subject to the GDPR’s EU-law successor regime), and the European-based PCA secretariat processed case materials in The Hague. Either of those facts, standing alone, was sufficient to engage the Regulation.

The UK GDPR. Following the United Kingdom’s withdrawal from the European Union, the United Kingdom retained the GDPR through the Data Protection Act 2018 and the Data Protection, Privacy and Electronic Communications (Amendments etc.) (EU Exit) Regulations 2019.7 The Tennant tribunal’s UK-based co-arbitrator was personally subject to the UK GDPR, regardless of NAFTA’s identity.

California’s CCPA, as amended by the CPRA. The California Consumer Privacy Act, codified at Cal. Civ. Code § 1798.100 et seq. and amended by the California Privacy Rights Act effective January 1, 2023, applies to businesses that collect or process the personal information of California residents and meet statutory thresholds.8 The investor’s California-resident counsel and California-resident expert witnesses participated in the Tennant proceeding. Their personal data, and the personal data of California-resident witnesses whose information was processed in connection with the case record, fell within the CCPA/CPRA’s scope. The California regime is enforced by the California Privacy Protection Agency, which has rule-making and administrative-fining authority that does not depend on the parties’ choice of arbitral seat.9 The tribunal’s confidentiality order was silent on the regime entirely.

Canada’s PIPEDA. The Personal Information Protection and Electronic Documents Act applies to private-sector processing of personal data in the course of commercial activities anywhere in Canada.10 The investor’s counsel was based in Toronto. The arbitration was seated in Toronto. The investor itself was a private-sector commercial party. PIPEDA applied.

Canada’s Privacy Act. The Privacy Act governs federal government institutions in their handling of personal data. 11 The Government of Canada was the respondent. The respondent’s processing of investor and witness personal data in connection with the proceeding fell squarely within the Privacy Act’s scope.

Singapore’s PDPA. The Personal Data Protection Act 2012 applies to organizations that collect, use, or disclose personal data in Singapore. 12 The presiding arbitrator was established in Singapore and practiced there throughout the proceeding. The presiding arbitrator’s processing of case materials was governed by the PDPA.

A single proceeding. Six mandatory regimes. The tribunal addressed none of them.

Why “Mandatory” Means What It Says

International arbitration has long accommodated mandatory norms. Tribunals have applied mandatory tax laws despite contrary party stipulations, recognized legal privilege as inviolable irrespective of the procedural code, and invalidated agreements tainted by corruption on public policy grounds. The structural logic is settled: certain obligations attach regardless of party choice because the obligations protect interests the parties cannot waive.

Data protection regimes belong on that list. The GDPR, the UK GDPR, the CCPA/CPRA, PIPEDA, the Canadian Privacy Act, and the Singapore PDPA each protect a fundamental right of natural persons whose personal data is processed in the course of arbitration. None of those regimes asks for the parties’ permission. None can be displaced by an arbitration agreement. None turns on whether the European Union, the United Kingdom, the State of California, Canada, or Singapore is a party to the underlying treaty. 13

The lex data is the doctrinal name for this overlay. 14 It is the mandatory, extraterritorial data protection framework that attaches to international arbitral proceedings through five digital connecting factors: the seat, the location of arbitrators, counsel, and institutional secretariats, the location of data subjects, the location of data storage and processing, and the location where the proceeding’s outputs will be used. Each connecting factor is independent. Each can engage a different regime. Where two or more attach simultaneously, the proceeding becomes subject to multiple regimes operating in parallel. This is the transnational simultaneity problem, and it is now a feature of every digitally active international arbitration.

What ARIHQ Now Adds: Data Privacy Breach as an Enforcement Risk

For the seven years after the Tennant ruling, the consequence of ignoring the lex data was, in practice, deferred. Tribunals could decline to perform the analysis without an immediately visible cost. ARIHQ changed that.

In ARIHQ v. Santé Québec, Justice Sheehan of the Quebec Superior Court annulled an arbitral award rendered August 8, 2025, by sole arbitrator Michel Jeanniot. The award was annulled on multiple grounds, four of which addressed the arbitrator’s undisclosed use of generative AI to produce hallucinated legal authorities. A fifth ground, less commented on but equally important for present purposes, addressed data sovereignty directly. The court observed that entering party information into an AI platform likely transferred personal data to servers outside Quebec and Canada, and that such a transfer would violate Quebec’s and Canada’s data privacy laws. 15 In the ARIHQ court’s framing, this is not an abstract compliance problem. It is a jurisdictional problem that affects the validity of the arbitral process itself.

The doctrinal step is significant. The ARIHQ court did not treat data protection law as parallel to the arbitration. It treated compliance with data protection law as part of what makes the arbitral process valid. Breach of data privacy law in connection with the proceeding, in ARIHQ’s analysis, undermines the integrity of the procedure and the validity of any resulting award.

That linkage matters far beyond Quebec. New York Convention Article V(2)(b) permits a recognition court to refuse enforcement of an award where recognition would be contrary to the public policy of the enforcing forum. Schrems II identifies data protection as constitutional public policy in the European Union. 16 Awards rendered in proceedings that disregarded mandatory data protection obligations, or that involved transfers of personal data in violation of the GDPR, the UK GDPR, the CCPA/CPRA, PIPEDA, the Canadian Privacy Act, or the Singapore PDPA, become structurally vulnerable at the recognition stage. The Lex Data working paper develops the analysis for three enforcement jurisdictions and concludes that the risk is no longer hypothetical. 17

Tennant defined the failure mode at the front end of the proceeding: the tribunal declines to identify which data protection regimes apply. ARIHQ defined the failure mode at the back end: the arbitrator breaches a data protection regime in connection with the proceeding, and the resulting award does not survive judicial review. The two cases connect through a single proposition: data protection law is integral to the validity of the arbitral process, and its breach is no longer an off-balance-sheet liability.

What Proactive Lex Data Governance Looks Like

The tribunal that has identified the applicable regimes at the first procedural meeting, allocated controller and processor roles transparently, addressed cross-border transfer mechanisms, conducted a Data Protection Impact Assessment where Article 35 requires it, and documented compliance, has built the record that defeats Article V(2)(b) at the enforcement stage. When California-resident participants are involved, the tribunal has identified and addressed the CCPA/CPRA threshold question. Where AI tools are used in the proceedings, whether by the arbitrators, counsel, or experts, the tribunal has documented the platform’s data-handling characteristics and ensured that personal data is not transferred to servers in a manner that breaches applicable regimes. The cost is an hour at the first session. The benefit is enforceability.

Why the Cybersecurity Protocol Is Necessary But Not Sufficient

The leading institutional instrument for digital governance in international arbitration is the ICCA-NYC Bar-CPR Cybersecurity Protocol for International Arbitration (2020, revised 2022).18 The Protocol normalized information security planning as a component of the arbitral procedure. Its contribution is substantial.

It is also incomplete. The Protocol is a cybersecurity instrument. It addresses the protection of case information against unauthorized access, use, disclosure, modification, and destruction. It does not address the affirmative obligations that data protection law imposes regardless of any security threat: the lawful basis for processing under GDPR Article 6; the controller and processor mapping under Article 26 and Article 28; the Data Protection Impact Assessment under Article 35; the cross-border transfer mechanisms under Articles 44 through 49; the consumer-rights mechanisms under the CCPA/CPRA; and the post-quantum security dimension that NIST FIPS 203, 204, and 205 (published August 2024) now adds to the Article 32 “appropriate to the risk” standard.19

The Lex Data working paper proposes five targeted enhancements to close the gap: a data-role mapping appendix, a DPIA decision tree, a transfer mechanism appendix, a multi-regime coordination matrix, and a post-proceeding data governance protocol that incorporates a documented plan for post-quantum re-encryption of any materials retained over a long time horizon. The enhancements are designed to be adopted by the institutional drafters of the Protocol’s next edition, not to compete with the Protocol.

How Tennant and ARIHQ Define the Arc

Tennant Energy and ARIHQ v. Santé Québec: the front and back covers are of the same volume. Tennant is the case in which a tribunal dismissed a claim under mandatory data protection law on grounds that did not survive analysis. ARIHQ is the case decided seven years later in which a court annulled an arbitral award in part because an arbitrator’s data-handling practices breached Quebec and Canadian data privacy laws and undermined the procedural integrity of the proceeding.

Both decisions belong to the same arc. Both turn on the same insight: the lex arbitri is necessary but no longer sufficient to govern the digital conduct of international arbitration. The lex data governs informational sovereignty. The companion lex AI framework governs the digital tribunal.20 The Digital Security Framework governs the security of the infrastructure. 21 Together with the lex arbitri, they constitute the Digital Procedural Constitution.

The Tennant tribunal could have closed the analytical gap in 2019 with one additional paragraph. The ARIHQ arbitrator could have avoided annulment in 2025 through disclosure and a different platform choice. Neither did. Their successors have run out of room to repeat the omission.

The TechCred Program

The American Bar Association’s TechCred program certifies digital competency for arbitrators and counsel. The program addresses, in three sessions, the core obligations the Lex Data framework identifies: identifying applicable data protection regimes through the digital connecting factors; allocating controller and processor responsibilities; integrating data protection into Procedural Order No. 1; and managing the cybersecurity, AI, and breach-response dimensions that intersect with the lex data.

The Fall 2026 TechCred program is now open for registration: https://events.americanbar.org/event/ee369916-3182-40d6-a1e9-5347aefeda2f/summary. The program is delivered live online and credentialed by New York Law School.

REGISTER NOW

A Final Word

The Tennant tribunal’s single-paragraph ruling is a useful artifact. It marks, with unusual clarity, the moment when the digital governance gap in international arbitration became visible. The proceeding processed multimillion-dollar claims across six mandatory data protection regimes simultaneously. The tribunal addressed none of them. The result was a regulatory vacuum that no party had requested and no law required.

The post-Tennant, post-ARIHQ tribunal has no excuse. The applicable regimes are identifiable. The connecting factors are objective. The procedural responses are now in the institutional literature and will shortly be consolidated into a Digital Governance Protocol designed to be dropped directly into Procedural Order No. 1.

When the next tribunal is asked at its first procedural meeting whether the GDPR, the CCPA/CPRA, or any other applicable regime applies, the answer is no longer one paragraph. It is an analysis. The lex data is the framework for that analysis. Performing it is now the baseline.

References

Appleton, Barry. Lex Data: Reconceptualizing the Law of the Digital Seat in International Arbitration (Working Paper, April 30, 2026), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6610079.

Appleton, Barry. Lex AI: Mandatory AI Governance and the Digital Tribunal in International Arbitration (Working Paper, April 30, 2026), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6610019.

Appleton, Barry. Arbitration Hacked: Toward a Unified Cybersecurity and Data Breach Framework for International Arbitral Proceedings (Working Paper, April 21, 2026), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6625218.

Appleton, Barry. ARIHQ v. Santé Québec: The Case That Changed Arbitrator AI Liability, TechCred Substack (May 4, 2026).

Association des ressources intermédiaires d’hébergement du Québec (ARIHQ) c. Santé Québec — Centre intégré universitaire de santé et de services sociaux du Centre-Sud-de-l’Île-de-Montréal, 2026 QCCS 1360 (Que. Sup. Ct. Apr. 22, 2026) (Sheehan, J.C.S.), https://canlii.ca/t/kkjtm.

California Consumer Privacy Act, Cal. Civ. Code §§ 1798.100–1798.199.100 (West 2024), as amended by the California Privacy Rights Act of 2020.

ICCA, New York City Bar Association & CPR Institute, Cybersecurity Protocol for International Arbitration (2022 ed., first published 2020), https://drs.cpradr.org/rules/protocols-guidelines/icca-nyc-bar-cybersecurities.

ICCA-IBA Joint Task Force on Data Protection in International Arbitration, Roadmap to Data Protection in International Arbitration, ICCA Reports No. 7 (2022), https://www.arbitration-icca.org/icca-reports-no-7-icca-iba-roadmap-data-protection-international-arbitration.

National Institute of Standards and Technology, Federal Information Processing Standards Publications 203, 204, and 205 (August 2024), https://csrc.nist.gov/projects/post-quantum-cryptography.

Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 (General Data Protection Regulation), 2016 O.J. (L 119) 1.

Prof. Barry Appleton, FCIArb, TechCred Faculty Director, ABA TechCred Program,
Co-Director and Distinguished Senior Fellow, New York Law School

This is the second piece in an ongoing series.

On November 3, 2025, the American Arbitration Association (AAA) and its international division, the International Centre for Dispute Resolution (ICDR), launched the AAA-ICDR AI Arbitrator. It is the first deployment of artificial intelligence in the adjudicative function of arbitration by a major international institution.1

The product is currently available for two-party, documents-only construction cases. It generates draft awards from case submissions, is trained on more than 1,500 prior construction awards, and is refined through expert-labeled examples and structured legal prompt libraries. The system was developed in collaboration with QuantumBlack, AI by McKinsey, and tested across more than 1,000 simulated cases.2 The AAA has announced plans to expand into insurance, commercial disputes, multi-party proceedings, and cross-border matters in subsequent phases.

The AI Arbitrator does not issue awards on its own. A human arbitrator, specifically trained for AI-led case reviews, revises if necessary, confirms, and signs. The Integra Ledger blockchain partnership provides tamper-proof authentication of the entire documentary record.3 In structural terms, the system is designed to satisfy the “human oversight” requirement of EU AI Act Article 14 and the non-delegation principles articulated in ARIHQ v. Santé Québec.4

Whether the architecture in fact does satisfy those obligations is the question this post examines. The answer matters because the AAA-ICDR AI Arbitrator is the first concrete test of whether mandatory AI governance can be reconciled with institutional efficiency. If it can, it sets the template for every other major institution. If it cannot, it sets the limit.

Why This Is a Lex AI Question

International arbitration is now subject to two direct AI governance instruments and a growing body of indirect regulatory architecture. The two direct instruments are the EU AI Act5 and ABA Formal Opinion 512.6 The indirect architecture includes the California Transparency in Frontier Artificial Intelligence Act (SB 53)7 and the New York Responsible AI Safety and Education Act (RAISE Act).8 None of these instruments is principally about arbitration. All of them now apply to it.

The Lex AI working paper proposes that this overlay constitutes a distinct mandatory framework on international arbitral proceedings, analogous to the lex data but coequal and independent.9 It is the law of the digital tribunal. Like the lex data, it attaches through objective connecting factors and cannot be displaced by party agreement.

The AAA-ICDR AI Arbitrator engages the Lex AI on every available connecting factor. It is a regulated AI system under the EU AI Act. It performs an arbitral function classified as high risk under Annex III, point 8(a). It is deployed by an institution whose case work routinely produces awards used in the European Union, which engages Article 2(1)(c)’s extraterritorial reach. It is used in connection with proceedings in which one or more arbitrators may be EU-established, which engages Article 2(1)(b). And its use by US-barred counsel is subject to ABA Formal Opinion 512.

The High-Risk Classification Is Not an Aspiration

The EU AI Act classifies AI systems intended for use by competent authorities in the administration of justice and democratic processes, including alternative dispute resolution, as high-risk. Recital 61 makes the legislative intent explicit. Annex III, point 8(a), is the operative provision.

The Stockholm Chamber of Commerce Guide to the Use of Artificial Intelligence in Cases Administered Under the SCC Rules (October 2024) acknowledges that the EU AI Act “qualifies as high-risk AI systems used by an Arbitral Tribunal ‘in researching and interpreting facts and the law and in applying the law to a concrete set of facts’”, citing Annex III, point 8(a) verbatim. 10 The Chartered Institute of Arbitrators Guideline on the Use of AI in Arbitration (March 2025, updated September 2025) similarly identifies AI use in arbitration as presenting risks to “enforceability of arbitral awards and other adverse implications for due process rights, the rule of law, the administration of justice, the credibility and legitimacy of arbitration”. 11 Both institutions thus acknowledge the high-risk character of AI in adjudicative contexts.

Neither institutional instrument provides the mandatory implementation mechanism that high-risk classification requires. The Lex AI framework supplies it through three mechanisms applicable at the first procedural meeting:

1. Risk-tier classification. Each AI system in the proceeding is characterized by EU AI Act tier: unacceptable-risk (prohibited), high-risk (mandatory compliance), limited-risk (transparency required), or minimal-risk. The characterization is documented in the procedural record.

2. Disclosure on a three-tier framework. Tier 1 administrative use (translation, format conversion) requires no disclosure. Tier 2 research assistance requires disclosure. Tier 3 substantive reasoning, including award drafting, requires disclosure, party consent, and verification. The framework is described in the Lex AI working paper and tested against the ARIHQ facts. 12

3. Platform qualification. A platform may be used in a proceeding only if it satisfies enterprise licensing terms that prohibit training on submitted data, that satisfy GDPR Article 28 processing-agreement requirements where applicable, and that offer the post-quantum cryptographic agility required by the lex data framework’s reading of Article 32. 13

The AAA-ICDR AI Arbitrator’s documented architecture suggests that it is designed with these requirements in mind: enterprise-grade infrastructure, four-step human-in-the-loop workflow, tamper-proof documentary authentication, and confined application to two-party documents-only construction cases. The next several phases of expansion will test how the architecture performs as scope grows.

Equality of Arms and Agentic AI

The EU AI Act’s transparency requirements reflect a deeper procedural justice concern that applies across the Lex AI framework: the equality of arms. 14 Agentic AI systems deployed to monitor proceedings, identify inconsistencies in opposing arguments, and suggest real-time procedural responses are now in use by certain well-resourced parties in high-value international arbitrations. A party whose counsel deploys such a system gains a real-time analytical advantage over a party whose counsel does not.

The Silicon Valley Arbitration and Mediation Center Guidelines on the Use of Artificial Intelligence in Arbitration (April 30, 2024) specifically recommend that tribunals consider whether asymmetric access to AI tools compromises procedural fairness and address the issue in Procedural Order No. 1. 15 Lex AI extends that recommendation into a mandatory framework: where the asymmetry is material, the tribunal’s duty to ensure procedural fairness now includes a duty to consider the AI architecture each party deploys, including agentic systems.

For institutional AI products like the AAA-ICDR AI Arbitrator, the equality-of-arms analysis takes a different form. When the institution itself supplies the AI, the question is not whether one party has access. It is whether the system has been disclosed, whether the parties have meaningfully consented to its use, and whether the human arbitrator’s review of the AI-generated draft is substantive rather than perfunctory.

The Hallucination Liability Chain

Five years. Five cases. Three jurisdictions. One unmistakable pattern. 16 In 2023, a New York federal court sanctioned counsel for submitting AI-generated, fabricated citations and then doubling down when challenged. 17 In 2024, the Second Circuit entered a disciplinary referral against an attorney who submitted a nonexistent ChatGPT-generated authority.18 In November 2025, the Qatar Financial Centre Civil and Commercial Court in Doha issued a landmark judgment addressing the same conduct. 19 In early 2026, the New York Appellate Division sanctioned counsel who continued filing AI-fabricated citations after being warned. 20In April 2025, a petitioner argued before a United States federal court that an arbitrator had outsourced the reasoning behind a twenty-nine-page award to artificial intelligence, presenting the first judicial test of whether AI-assisted award drafting constitutes grounds for vacatur under the Federal Arbitration Act. 21

In April 2026, the chain reached the developer. Nippon Life Insurance Company of America v. OpenAI Foundation, filed March 4, 2026, in the Northern District of Illinois, is the first tort action naming an AI developer as a defendant in litigation-conduct claims. 22In April 2026, the Quebec Superior Court annulled an arbitral award in ARIHQ v. Santé Québec because the arbitrator’s undisclosed AI use produced hallucinated authorities. 23

The chain runs counsel, arbitrator, and developer. Mata, Park, Sheppard, Deutsche Bank, LaPaglia, ARIHQ, Nippon Life. Six years. Each case extends the Lex AI framework from where the prior case left off. Each case clarifies that the duty of verification is now a hard rule.

Why the Voluntary Framework Cannot Carry the Load

The SCC AI Guide is candid about its scope: it provides “flexible guidance to participants in an SCC case without imposing specific obligations”. 24 The CIArb AI Guideline is equally explicit: it is “intended for use in conjunction with, and does not supersede, any applicable laws, regulations or policies”, and where it is “incompatible with any agreement of the parties, and or applicable mandatory rules and regulations, the latter shall of course prevail”. 25

Both institutional instruments thus subordinate themselves to mandatory law. Neither constitutes the mandatory framework it acknowledges. Where the EU AI Act’s high-risk obligations apply, the obligations apply whether or not a tribunal has consulted the SCC Guide or the CIArb Guideline. Where ABA Formal Opinion 512 applies to a US-barred lawyer, the Opinion applies whether or not counsel has consulted any guideline at all.

The Lex AI working paper therefore proposes an enhanced model procedural clause designed to operationalize the mandatory obligations at the first procedural meeting. The clause is structured around three pillars: applicable regimes (mapped through connecting factors); risk-tier characterization (with documentation); and disclosure, verification, and oversight obligations (calibrated to the tier). 26

A tribunal that adopts the CIArb Appendix B Option 1 short-form order, which simply directs participants to “be guided by the CIArb AI Guideline in the conduct of the proceedings”, has not discharged its EU AI Act compliance obligations for high-risk AI systems. A tribunal that adopts the Lex AI model clause has.

The TechCred Program

The American Bar Association’s TechCred program is the first internationally accredited certification in AI, digital innovation, and technology dispute resolution. The curriculum addresses the Lex AI framework directly: the EU AI Act risk-tier system, ABA Formal Opinion 512 obligations, the equality-of-arms problem, hallucination prevention, and the integration of AI governance into Procedural Order No. 1.

The Fall 2026 TechCred program is open for registration: https://events.americanbar.org/event/ee369916-3182-40d6-a1e9-5347aefeda2f/summary. The program is delivered live online, accredited by New York Law School, and is the practitioner companion to the working papers in this series.

REGISTER NOW

A Final Word

The AAA-ICDR AI Arbitrator is the institutional product that the arbitration community has been preparing for since the first AI hallucination case reached a court. It is also the most carefully structured deployment of AI in the adjudicative function to date. Whether its architecture survives the first New York Convention enforcement challenge, the first ABA disciplinary inquiry, and the first EU Member State court asked to recognize an AAA-ICDR AI-assisted award is the question the next two years will answer.

What is no longer in doubt is the framework within which the answer will be assessed. The EU AI Act applies. ABA Formal Opinion 512 applies. The Lex AI connecting factors will be tested in real cases. The ARIHQ line is now the precedent that gives all of this judicial flesh.

When the next AAA-ICDR AI Arbitrator award is challenged, the case file will not contain a single paragraph dismissing the relevant AI governance regime. It will contain a record built to show that every regime was identified, every obligation was performed, and every disclosure was made. That record is the institutional product, more than any algorithm.

References

Appleton, Barry. Lex AI: Mandatory AI Governance and the Digital Tribunal in International Arbitration (Working Paper, April 30, 2026), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6610019.

Appleton, Barry. Lex Data: Reconceptualizing the Law of the Digital Seat in International Arbitration (Working Paper, April 3, 2026), https://ssrn.com/abstract=6610079.

Appleton, Barry. Artificial Intelligence and International Arbitration: A Comprehensive Listing of Governing Laws, Institutional Rules, Professional Responsibility Standards, and Guidance (Working Paper, May 4, 2026), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6631182.

American Arbitration Association — International Centre for Dispute Resolution, Guidance on Arbitrators’ Use of AI Tools (March 2025), https://go.adr.org/rs/294-SFS-516/images/2025_AAAICDR%20Guidance%20on%20Arbitrators%20Use%20of%20AI%20Tools.pdf?version=0.

Chartered Institute of Arbitrators, Guideline on the Use of AI in Arbitration (March 2025, updated September 2025), https://www.ciarb.org/media/bpndtcgu/guideline-on-the-use-of-ai-in-arbitration_updated-sept-2025.pdf.

Stockholm Chamber of Commerce Arbitration Institute, Guide to the Use of Artificial Intelligence in Cases Administered Under the SCC Rules (October 2024).

Silicon Valley Arbitration and Mediation Center, Guidelines on the Use of Artificial Intelligence in Arbitration (April 30, 2024), https://svamc.org/wp-content/uploads/SVAMC-AI-Guidelines-First-Edition.pdf.

Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act).

American Bar Association Standing Committee on Ethics and Professional Responsibility, Formal Opinion 512: Generative Artificial Intelligence Tools (July 29, 2024).

Something happened in a Quebec courtroom last month that arbitrators everywhere need to know about — and it just landed on the TechCred Substack.

On April 22, 2026, in ARIHQ v. Santé Québec, the Quebec Superior Court became the first court anywhere in the world to annul an arbitral award because the arbitrator had used AI to draft it without telling the parties. Five authorities cited in the award turned out to be hallucinations. Articles that were never written. Court of Appeal decisions that do not exist. An arbitral precedent that Quebec’s official legal service confirmed was pure invention.

The arbitration community has been waiting for this case since LaPaglia v. Valve was filed last year and dismissed on jurisdictional grounds before the merits could be reached. ARIHQ reaches them. And the principles Judge Sheehan set out — on disclosure, on the personal nature of the arbitral mandate, on the limits of delegation to a machine — apply far beyond Quebec.

Prof. Barry Appleton has written the first detailed analysis of the decision and what it means for international arbitration. He works through the four doctrines the court applied, traces their analogs under the FAA, the English Arbitration Act 2025, the Singapore IAA, the Hong Kong Ordinance, and the EU AI Act, and shows how the decision aligns with the AAA-ICDR, SCC, VIAC, and CIArb guidance issued over the past year.

If you read one piece on AI and arbitrator liability this year, make it this one. And if you want more, there are two connected pieces to follow — publishing on May 11 and May 12

What’s next

ARIHQ is going to shape the conversation for the rest of 2026, and TechCred is going to be at the center of it.

• Tuesday, May 12 at 4:30 PM — Prof. Appleton will present the case at the *AI in the Seat of Justice* panel at the ABA International Law Section Annual Conference in Washington, DC. If you are at the conference, come find us.

• November 19, 2026 — ARIHQ will be a featured topic at the 90-minute-long TechCred 2026 annual recertification update, alongside the AAA-ICDR AI Arbitrator, the Digital Omnibus, and the new wave of institutional rule changes that are arriving in response to cases just like this one.

• October 22,23 and 29, 2026 - For your friends who did not take TechCred in 2025, we have our dates set for TechCred 2026.

REGISTER NOW

This is exactly why this TechCred community exists. Forward this to a colleague who needs to see it, and we will see you in DC and in November.

Warmly,

Katja Maas
TechCred, Team / Maas Creatives

By Prof Barry Appleton, FCIArb, TechCred
Faculty Director, ABA TechCred Program, Co-Director and Distinguished Senior Fellow, New York Law School

This is the first piece in a series of three. Next on 05/11 and 05/12

On April 22, 2026, the Quebec Superior Court handed down a decision that rewrites the rulebook for artificial intelligence in arbitration. In Association des ressources intermédiaires d’hébergement du Québec (ARIHQ) c.Santé Québec - Centre intégré universitaire de santé et de services sociaux du Centre-Sud-de-l’Île-de-Montréal, Justice Martin Sheehan became the first judge globally to annul an arbitral award based on an arbitrator’s undisclosed use of AI to generate substantive reasoning supported by hallucinated authorities.

The decision answers a question that has haunted the international arbitration community since LaPaglia v. Valve Corp., No. 3:25-cv-00833-RBM-DDL (S.D. Cal., petition filed Apr. 8, 2025; dismissed Dec. 9, 2025 for lack of subject matter jurisdiction), which was filed last year: what happens when an arbitrator delegates award drafting to AI? LaPaglia never reached the merits. ARIHQ did.

What happened in the ARIHQ case?

The arbitration concerned a contractual dispute between Quebec’s health authority and an association representing residential care facilities. The sole arbitrator, Michel Jeanniot, issued an award on August 8, 2025, addressing limitation periods under Quebec law. The award cited five authorities. All five were fabricated by artificial intelligence.

These “authorities” do not exist:

• An article purportedly authored by Professor Frédéric Bachand titled “Prescription et déchéance: frontières mouvantes et enjeux pratiques,” supposedly published in a 2016 Barreau du Québec collection. Professor Bachand of McGill University is a real and respected scholar. He did not write this article.

• A purported Quebec Court of Appeal decision cited as Ville de Montréal c. Syndicat des cols bleus, 2005 QCCA 591.

• A purported Quebec Superior Court decision cited as Groleau et Groupe Pages Jaunes Cie, 2011 QCCS 5386.

• A purported Quebec Court of Appeal decision cited as Tremblay c. Commission scolaire de la Jonquière, 2002 CanLII 24357 (QCCA).

• A purported arbitral precedent cited as Arbitrage CHU Ste-Justine (D.T.E. 2018-30). SOQUIJ, Quebec’s official legal information service, confirmed this precedent does not exist.

The neutral citations for the court “decisions” lead to entirely different, unrelated cases. No warning appears here without cause: arbitrators who accidentally cite these fabricated authorities risk perpetuating AI-generated hallucinations.

Why the Court Annulled the Award

Justice Sheehan analyzed the problem through four interlocking doctrines of Quebec arbitration law:

• The personal mandate doctrine. An arbitrator’s authority derives from party consent. That consent is to the arbitrator’s personal exercise of judgment, not to a machine’s. Delegation to AI substitutes a different decision-maker than the one the parties selected.

• The reasoned award doctrine. Quebec law requires arbitrators to provide reasoning. Reasoning copied from AI output without independent verification is not the arbitrator’s reasoning. The court invoked the Supreme Court of Canada’s observation in Baker v. Canada (Minister of Citizenship and Immigration), [1999] 2 SCR 817, that the process of writing reasons can in itself guarantee a better decision.

• The secret of deliberation. Quebec arbitrators have an affirmative duty not to communicate with third parties about a matter under advisement. Entering case facts into an AI platform is communication with a third party. The AI system becomes a consultant on the merits, which is prohibited during deliberations.

• The non-delegation doctrine. The Supreme Court of Canada established in Therrien (Re), 2001 SCC 35, that “the body to which the exercise of a power is attributed under its enabling statute must exercise it itself.” An arbitrator cannot delegate adjudicative judgment to a machine.

The court also addressed data sovereignty. Inputting party information into an AI platform likely transfers personal data to servers outside Quebec and Canada. This violates Quebec’s and Canada’s data privacy laws. For arbitrators, this is not an abstract compliance problem. It is a jurisdictional problem that affects the validity of the arbitral process itself.

Why This Matters Beyond Quebec

The ARIHQ decision applies everywhere, not just in Quebec. Here’s why.

In the United States. The Federal Arbitration Act allows vacatur of awards procured through “evident partiality,” “misbehavior,” or where “the arbitrators exceeded their powers.” The four doctrines that Justice Sheehan applied have plausible analogs in FAA section 10(a)(4) (arbitrators exceeded their powers), although the merits question remains open after LaPaglia’s dismissal on subject matter jurisdiction grounds. The AAA-ICDR Guidance on Arbitrators’ Use of AI Tools (March 2025) already requires arbitrators to disclose any use of AI that materially affects the arbitration process or reasoning. ARIHQ now provides a judicial precedent showing what happens when arbitrators fail to do so.

In Commonwealth jurisdictions. The English Arbitration Act 2025, the Singapore International Arbitration Act, the Hong Kong Arbitration Ordinance, and Canada’s provincial arbitration statutes all recognize duties of independence, impartiality, and procedural fairness. An arbitrator’s undisclosed use of AI to generate substantive reasoning violates these duties. The Chartered Institute of Arbitrators issued its Guideline on the Use of AI in Arbitration (March 2025, updated September 2025), stating that “arbitral functions should not be delegated to AI.”

Under risk-based AI regulation. The EU AI Act (Regulation (EU) 2024/1689) classifies AI systems used “in researching and interpreting facts and the law and in applying the law to a concrete set of facts” as high-risk (Annex III, point 8(a)). The high-risk obligations under Annex III take effect on August 2, 2026, subject to ongoing trilogue negotiations on the European Commission’s Digital Omnibus on AI proposal. California’s Transparency in Frontier Artificial Intelligence Act (TFAIA, SB 53) and New York’s Responsible AI Safety and Education Act (RAISE Act) regulate frontier AI developers. These regulatory frameworks establish that AI-assisted adjudication is an inherently high-stakes activity requiring disclosure, verification, and human oversight.

Through institutional arbitration rules. The ARIHQ decision aligns with emerging institutional guidance:

• The AAA-ICDR Guidance (March 2025) requires disclosure when AI use “materially impacts the arbitration process or the reasoning underlying their decisions.”

• The Stockholm Chamber of Commerce Guide to the Use of Artificial Intelligence in Cases Administered Under the SCC Rules (October 2024) warns that AI use raises questions about the independent judgment of the arbitrator.

• The Vienna International Arbitral Centre Note on the Use of Artificial Intelligence in Arbitration Proceedings (April 2025) provides that AI may not substitute for the arbitrator’s reasoning.

• The Silicon Valley Arbitration and Mediation Center Guidelines on the Use of Artificial Intelligence in Arbitration (April 2024) include Guideline 6: “Arbitral decision-making cannot be delegated to AI.”

• The China International Economic and Trade Arbitration Commission issued its Provisional Guidelines on the Use of Artificial Intelligence Technology in Arbitration (Trial) effective July 18, 2025, requiring disclosure and prohibiting reliance on unverified AI outputs.

The International Chamber of Commerce revised its Rules of Arbitration (effective June 1, 2026), but these revisions do not yet address arbitrators’ use of AI. The London Court of International Arbitration has not yet issued AI-specific guidance. Both arbitration entities have rules committees looking at these issues at this time. The gap matters. Institutional rules without AI provisions leave arbitrators, and the awards they render, exposed to the challenges ARIHQ now makes possible.

What ARIHQ Tells Arbitrators

The ARIHQ decision establishes five principles that apply to every arbitrator in every jurisdiction:

• Disclosure is foundational. Any use of AI that affects the integrity of the decision-making process must be disclosed to the parties. Disclosure gives parties the opportunity to consent, object, or require verification procedures. Without disclosure, parties cannot exercise informed consent.

• Verification is non-delegable. An arbitrator who uses AI to draft portions of an award must independently verify every statement of fact, every citation, and every element of reasoning. Verification cannot be delegated to the AI system itself. Asking the AI to “check its work” is not verification.

• Materiality has tiers. Not all AI use requires disclosure. AI use for administrative tasks (calendar management, email drafting, translation of submissions) does not materially affect adjudication. AI use for research assistance may require disclosure depending on how the results are used. AI used to draft substantive reasoning in the award always requires disclosure.

• Deliberation is personal. During deliberation, an arbitrator must not communicate with third parties about the merits of the case. Entering case facts into an AI platform is arguably communication with a third party. This prohibition applies during deliberations, even if the AI tool has been disclosed to the parties and they consented to its use.

• Some violations are non-waivable. Party consent does not cure all procedural defects. If an arbitrator delegates the decision-making function itself to AI, meaning the arbitrator reviews but does not independently reason through the AI’s output, the award may be annulled even if the parties consented to AI use. The arbitrator’s personal exercise of judgment is mandatory.

The TechCred Program

The American Bar Association’s TechCred program certifies digital competency for arbitrators. The first TechCred class graduated in 2025 and received the competency credential. The ARIHQ decision confirms what the TechCred program has been teaching since 2025: arbitrators must understand AI well enough to know when they are delegating judgment rather than using a tool. They must be able to evaluate AI outputs, detect hallucinations, and implement verification procedures. And they must know how to communicate the use of AI to parties in procedural orders and disclosures.

The next TechCred three-day international arbitration competency program takes place in October and November 2026. The program, delivered live but online, is the first internationally accredited certification in AI, digital innovation, and technology dispute resolution, presented by the ABA International Law Section. Registration opens at https://events.americanbar.org/event/ee369916-3182-40d6-a1e9-5347aefeda2f/summary

REGISTER NOW

A Final Word

The ARIHQ case is the first judicial decision globally to annul an arbitral award on the grounds of an arbitrator’s use of AI. It will not be the last. Damien Charlotin’s AI Hallucination Cases database catalogs 1,394 AI hallucination cases globally as of May 1, 2026. ARIHQ is the only case in that database addressing arbitrator use. But the pattern is clear: AI hallucination liability has moved from counsel (Mata v. Avianca, Park v. Kim, Sheppard v. Jillion LLC, Deutsche Bank v. LeTennier) to arbitrators (ARIHQ), and it will soon reach developers (Nippon Life Insurance Co. v. OpenAI Foundation, filed March 4, 2026, remains pending in the Northern District of Illinois).

For arbitrators, the lesson is simple: disclosure, verification, and human judgment are not optional. They are the baseline for lawful AI use in adjudication. The ARIHQ court did not hold that arbitrators cannot use AI. The court held that arbitrators cannot delegate judgment to AI, cannot hide AI use from the parties, and cannot rely on unverified AI outputs in their reasoning. Those principles apply in every seat of arbitration under every arbitration law.

References

Association des ressources intermédiaires d’hébergement du Québec (ARIHQ) c. Santé Québec - Centre intégré universitaire de santé et de services sociaux du Centre-Sud-de-l’Île-de-Montréal, 2026 QCCS 1360 (Que. Sup. Ct. Apr. 22, 2026) (Sheehan, J.C.S.), https://canlii.ca/t/kkjtm

Therrien (Re), 2001 SCC 35, [2001] 2 SCR 3, https://canlii.ca/t/521j

Baker v. Canada (Minister of Citizenship and Immigration), [1999] 2 SCR 817, 1999 CanLII 699 (SCC), https://canlii.ca/t/1fqlk

LaPaglia v. Valve Corp., No. 3:25-cv-00833-RBM-DDL (S.D. Cal., petition filed Apr. 8, 2025; dismissed Dec. 9, 2025 for lack of subject matter jurisdiction), https://dockets.justia.com/docket/california/casdce/3:2025cv00833/810531

Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023) (Castel, J.)

AAA-ICDR, Guidance on Arbitrators’ Use of AI Tools (March 2025), https://go.adr.org/rs/294-SFS-516/images/2025_AAA-ICDR%20Guidance%20on%20Arbitrators%20Use%20of%20AI%20Tools%20%282%29.pdf?version=0

Chartered Institute of Arbitrators, Guideline on the Use of AI in Arbitration (March 2025, updated September 2025), https://www.ciarb.org/media/bpndtcgu/guideline-on-the-use-of-ai-in-arbitration_updated-sept-2025.pdf

Stockholm Chamber of Commerce Arbitration Institute, Guide to the Use of Artificial Intelligence in Cases Administered Under the SCC Rules (October 2024), https://sccarbitrationinstitute.se/wp-content/uploads/2024/12/scc_guide_to_the_use_of_artificial_intelligence_in_cases_administered_under_the_scc_rules-1.pdf

Vienna International Arbitral Centre, Note on the Use of Artificial Intelligence in Arbitration Proceedings (April 2025), https://www.viac.eu/wp-content/uploads/2025/04/VIAC-Note-on-AI.pdf

Silicon Valley Arbitration and Mediation Center, Guidelines on the Use of Artificial Intelligence in Arbitration (April 30, 2024), https://svamc.org/wp-content/uploads/SVAMC-AI-Guidelines-First-Edition.pdf

China International Economic and Trade Arbitration Commission, Provisional Guidelines on the Use of Artificial Intelligence Technology in Arbitration (Trial) (effective July 18, 2025)

International Chamber of Commerce, ICC Rules of Arbitration (2026), effective June 1, 2026, https://iccwbo.org/dispute-resolution-services/arbitration/rules-of-arbitration

Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act), Annex III, point 8(a) (administration of justice and democratic processes)

Damien Charlotin, AI Hallucination Cases, https://www.damiencharlotin.com/hallucinations/ (last visited May 4, 2026)

ABA TechCred Program, https://events.americanbar.org/event/ee369916-3182-40d6-a1e9-5347aefeda2f/summary

By Tejas Gupta (2025), SSRN

Abstract

The rapid advancement of digital technologies has fundamentally transformed the landscape of dispute resolution worldwide. This research paper explores the integration of Online Dispute Resolution (ODR) within the broader legal ecosystem, particularly in the Indian context, and examines the dual nature of digital disruption-its transformative potential as well as the inherent risks. It traces the evolution from traditional methods such as litigation and Alternative Dispute Resolution (ADR) to ODR systems powered by artificial intelligence, blockchain, and smart contracts.

The paper highlights the opportunities presented by ODR, including improved accessibility, cost and time efficiency, transparency, scalability, and global legal reform initiatives like UNCITRAL’s technical notes and India’s e-Courts Project. At the same time, it identifies challenges such as legal recognition gaps, jurisdictional complexities, cybersecurity risks, digital illiteracy, and regulatory fragmentation. Through comparative analysis of ODR models in the United States, European Union, and China, the paper outlines best practices and valuable lessons for India. It also investigates the emerging role of smart courts, predictive justice, and the importance of training for legal professionals. The final sections provide actionable recommendations on legal reform, user-centric platform design, and institutional collaboration between the judiciary, academia, and the tech sector. The paper concludes by advocating for a balanced and ethically grounded approach to digital dispute resolution—one that upholds fundamental principles of justice while embracing innovation through adaptive and inclusive governance.

Keywords: Online Dispute Resolution (ODR), Alternative Dispute Resolution (ADR), Digital Technologies, Dispute Resolution Worldwide, Evolution from Traditional Methods Such as Litigation and Alternative Dispute Resolution (ADR) to ODR Systems Powered by Artificial Intelligence, Blockchain, and Smart Contracts., Rapid Advancement of Digital Technologies has Fundamentally Transformed the Landscape of Dispute Resolution Worldwide.

Suggested Citation:

Gupta, Tejas, Dispute Resolution in a Digital Age: Challenges and Opportunities (July 04, 2025). Available at SSRN: https://ssrn.com/abstract=5339274 or http://dx.doi.org/10.2139/ssrn.5339274

READ FULL PAPER

TechCred Notes

• Strong fit for our Ethics, Consent, and Enforceability in Digital Disputes (TechCred Session 3).

• It complements Helberger’s theoretical “technology courts” paper by offering a practical, jurisdiction-specific, and comparative look at ODR.

• It’s more policy-oriented, making it valuable for practitioners, policymakers, and arbitrators looking at applied digital dispute resolution frameworks.

  
  
  

.

Bias is one of those words that arrives with its own weather. People hear it and brace for a storm: ideological bias, cultural bias, political bias. But in the algorithmic age, “bias” is something both simpler and stranger. It is the tilt that emerges whenever we ask a machine to compress the world into patterns we can use.

Bias isn’t a flaw. It’s a property. Everything that filters reality—eyes, memory, law, language—bends it a little. But AI makes that bending rigid.

For arbitrators navigating disputes that increasingly involve AI-generated evidence, AI-assisted analysis, or questions about algorithmic decision-making, understanding how models bend reality is no longer optional. It is foundational literacy. And like any literacy, it begins with learning to see what is already there.

What Bias Actually Means in AI

In AI, bias is not an opinion. It’s a structural feature. A statistical lean. A tendency wired into a model by what it has seen, what it has ignored, and what its designers have shaped it to avoid.

Three categories matter most to arbitrators:

Data Bias: The Inherited Absence

AI learns from human traces: text, legal documents, case law, news archives, academic papers. If parts of reality are missing from that archive, the model inherits the absence. If one jurisdiction, language, or legal tradition is over-represented, the model inherits the surplus.

Example in practice: An arbitrator using AI to summarize case law asks for precedents on force majeure in construction contracts. The model returns 47 cases—46 from common law jurisdictions, one from civil law. Not because civil law cases don’t exist, but because English-language legal databases dominate the training data. The bias isn’t malicious; it’s statistical. But to an arbitrator adjudicating an international construction dispute under Swiss law, the imbalance could be dispositive.

Research by Bommasani et al. (2021) in “On the Opportunities and Risks of Foundation Models” documents how training data composition directly shapes model outputs, with English-language content representing approximately 92% of training data in major models despite English speakers comprising only 17% of the global population.

Design Bias: The Philosophical Inheritance

Models don’t just absorb data. They inherit the philosophies of their makers. Every model has an internal logic—helpful, cautious, experimental, fast, bold, conservative. Safety tuning, reinforcement learning, guardrails: these are design decisions that shape the personality of the system.

Example in practice: An arbitrator asks two different models to assess whether certain communications constitute an anticipatory breach. Model A hedges extensively, noting seventeen reasons the question cannot be definitively answered without additional context. Model B provides a confident three-paragraph analysis. Neither is “wrong”—they’ve been tuned for different risk profiles. Model A was trained to avoid overconfidence; Model B to minimize user friction. The arbitrator who doesn’t recognize this difference might mistake caution for thoroughness or confidence for competence.

Interaction Bias: The Dance Between User and System

AI is unusually impressionable. Your phrasing, tone, order of questions, or even your uncertainty can nudge a model into a particular shape. Each prompt is a small gravitational pull.

Example in practice: An arbitrator asks: “Can you help me understand the weaknesses in the claimant’s position?” versus “Can you analyze the strengths and weaknesses of both parties’ positions?” The first prompt creates an adversarial frame; the model will search harder for flaws. The second invites balance. Same facts, same model, radically different analytical orientation.

Perez et al. (2022) in “Discovering Language Model Behaviors with Model-Written Evaluations” demonstrate that prompt framing can shift model outputs by up to 40% on factual accuracy tasks, with leading questions producing systematically different responses than neutral queries.

How Bias Manifests in Arbitration Contexts

Bias doesn’t announce itself. It appears in the seams. For arbitrators, five patterns matter most:

Jurisdictional Defaults

Models trained primarily on U.S. and EU legal materials may reflexively apply common law reasoning even when analyzing civil law questions. Ask about “discovery” and the model assumes American-style depositions unless explicitly redirected.

Temporal Compression

Models flatten legal evolution. Ask about the enforceability of arbitration agreements and the model may synthesize cases from 1995 and 2023 without clearly marking the doctrinal shift that occurred in between.

Consensus Bias

Models gravitate toward majority positions. On contested questions — say, the proper interpretation of Article 28 of the UNCITRAL Model Law — the model will favor the interpretation that appears most frequently in its training data, which may or may not reflect the best reasoned position or the position adopted in the relevant jurisdiction.

Hallucination Patterns

Different models fill knowledge gaps differently. A bold model will confidently invent case citations. A cautious model will refuse to answer. Both are forms of bias, one toward saying too much, the other toward saying too little. For an arbitrator relying on AI research assistance, the difference is critical.

Maynez et al. (2020) in “On Faithfulness and Factuality in Abstractive Summarization” found that 70% of neural abstractive summaries contained factual inconsistencies, with legal and technical domains showing higher error rates than general text.

Cultural Translation Failures

Legal concepts don’t translate cleanly. “Good faith” in American contract law is not bonne foi in French law is not Treu und Glauben in German law. Models trained to find equivalences will compress these distinctions unless actively prevented from doing so.

A Brief Aside: Cutting on the Bias

In dressmaking, cutting on the bias means slicing the fabric diagonally, at forty-five degrees to the weave. Do this, and something curious happens. The cloth gains stretch. It drapes differently. It moves with the body instead of resisting it. The same material, the same fibers, the same threads but the angle changes everything.

AI bias works a bit like that.

A model is made of data the way a fabric is made of threads. Straight-grain answers are the ones that follow the warp and weft of the training set. They’re sturdy, predictable, and occasionally stiff. But tilt the question, twist the context, or introduce a structural assumption and suddenly the model stretches. It drapes. It behaves differently.

Every model has its own “grain,” its own hidden geometry. Some fall with a clean line. Some pull unexpectedly at the seams. Some cling; some glide.

And just as an experienced seamstress learns to feel the grain through her fingertips, AI-literate arbitrators learn to sense where a model has stretch and where it has none. You learn to predict how it will fall. You learn when a straight cut will hold its shape and when cutting on the bias will give you something fluid.

Bias, in both senses, is not about right or wrong. It's about understanding the fabric you're working with. Once you understand it, you know what to expect.

The Institutional Architectures of Bias

Different organizations optimize for different values. These institutional priorities echo through every output.

OpenAI (GPT-4, GPT-o1)

Philosophy: Clarity, broad accessibility, moderate risk tolerance
Strength: Structured analytical reasoning, clear explanations
Bias pattern: Gravitates toward consensus positions, hedges on politically contested questions
Arbitration relevance: Reliable for procedural analysis, less useful for novel jurisdictional questions

Anthropic (Claude)

Philosophy: Constitutional AI, ethical awareness, transparency
Strength: Nuanced reasoning on ambiguous questions, explicit about limitations
Bias pattern: Over-cautious on questions involving harm, may refuse helpful analysis
Arbitration relevance: Excellent for multi-jurisdictional analysis, may need coaxing on adversarial content

Meta (Llama)

Philosophy: Open source, developer control, minimal restrictions
Strength: Adaptable, can be fine-tuned for specialized domains
Bias pattern: Highly variable depending on fine-tuning; less corporate safety filtering
Arbitration relevance: Best for specialized deployments; requires careful evaluation

Google (Gemini)

Philosophy: Multi-modal integration, enterprise focus
Strength: Handles complex document analysis across formats
Bias pattern: Optimized for business contexts, less granular legal reasoning
Arbitration relevance: Useful for document review, less reliable for doctrinal analysis

[Note These are general tendencies and fine-tuning, version updates, and deployment contexts can affect behavior.]

When you choose a model, you’re not choosing a tool. You’re choosing a worldview, a set of institutional priorities, and a philosophy of risk. An arbitrator who doesn’t understand this distinction is flying blind.

Practical Techniques for Working with Bias

Bias is not a dragon to be slain. It is a landscape to learn to navigate. Here are specific techniques arbitrators can use:

Technique 1: Prompt for Provenance

Don’t ask: “What does the case law say about waiver of arbitration rights?”

Ask: “What does the case law in [specific jurisdiction] say about waiver of arbitration rights? Please distinguish between pre-2015 and post-2015 cases, and identify any circuit splits or doctrinal shifts.”

This prompt forces the model to make its assumptions visible.

Technique 2: Adversarial Prompting

After receiving an initial analysis, ask:

• “What would be the strongest counterargument to this position?”

• “What assumptions are you making that might not hold in [civil law/common law/hybrid] systems?”

• “What edge cases or exceptions did you not mention?”

• “If you were representing the opposing party, how would you attack this reasoning?”

Welleck et al. (2023) in “Generating Sequences by Learning to Self-Correct” show that adversarial follow-up prompts reduce factual errors by 35% and increase reasoning depth in legal analysis tasks.

Technique 3: Multi-Model Triangulation

Use different models as checks on each other. Where they agree, you likely have consensus positions. Where they diverge, you’ve found the contested ground—which is often exactly what matters most in arbitration.

Technique 4: Explicit Framing

Begin prompts with: “You are an expert in [specific legal domain]. Your training data may over-represent [jurisdiction/legal tradition]. Be especially careful about...”

This metacognitive framing activates different parts of the model’s knowledge graph.

Technique 5: Few-Shot Examples

Provide 2-3 examples of the kind of analysis you want before asking your actual question. Models are remarkably good at pattern matching. Show them the pattern you need.

Example:

Here's an example of the level of detail I need:

Question: Is an arbitration clause in a franchise agreement enforceable if the franchisee is an individual?
Good answer: Under Swiss law, yes, but the clause must meet additional transparency requirements under Article 179(1) CO...

Now please answer this question: [your actual question]

Technique 6: Document Model Provenance

In any written decision or legal memorandum that relies on AI assistance, maintain a record of:

• Which model and version you used

• The specific prompts you provided

• When the analysis was generated

• What verification steps you took

This isn’t just good practice—as AI-assisted analysis becomes more common in arbitration, it will likely become a disclosure obligation.

Why This Matters for Arbitrators

Three scenarios make AI bias particularly salient in arbitration:

Scenario 1: Evaluating AI-Generated Evidence

A party submits contract language they claim was “drafted by AI” or damage calculations “verified by AI analysis.” An arbitrator must assess not just whether the output is accurate, but whether the AI’s biases might have shaped the result in legally significant ways.

Scenario 2: Using AI as Research Assistant

You’re deciding a complex jurisdictional question involving the interaction of three different national laws. You use AI to survey the relevant precedents. The model’s jurisdictional bias becomes your jurisdictional bias—unless you actively correct for it.

Scenario 3: Assessing Algorithmic Decision-Making

The dispute itself involves an algorithm: credit scoring, employment screening, pricing systems. To evaluate whether the algorithm was discriminatory or violated contractual obligations, you must understand how algorithmic bias works—not in theory, but in practice.

The Emerging Standards

The legal profession is beginning to develop standards for AI use. The ABA’s Resolution 604 (2024) on Generative AI emphasizes the duty of technological competence. Several jurisdictions now require disclosure of AI assistance in court filings.

For arbitrators, the standards are less formal but no less important:

Due Diligence: Understand the capabilities and limitations of any AI tool you use.

Verification: Never rely solely on AI-generated legal research or analysis. Verify citations, check primary sources, and confirm jurisdictional applicability.

Transparency: Consider whether AI assistance should be disclosed to the parties, particularly if it substantially shapes your analysis.

Independence: Ensure AI use enhances rather than replaces independent judgment. The arbitrator’s role is irreducibly human: weighing credibility, assessing context, applying judgment to ambiguous facts.

Conclusion: Discernment in the Algorithmic Age

AI will not replace arbitrators. But arbitrators who understand how models bend reality will think more clearly inside complexity than those who assume the machine is neutral.

In the end, bias isn’t the enemy. The unseen is the enemy.

Once you can see the tilt, you can steer around it.
Once you can name the pattern, you can break it.
Once you understand the worldview of the tool, you can decide when to trust it and when to hold the pen yourself.

The algorithmic age doesn’t demand perfection. It demands discernment.

And discernment, in its oldest sense, means this: the ability to tell one thing from another, to see the contours that others miss, and to navigate complexity with a mind awake to its own shadows.

That is the real skill of this era.
And it belongs to anyone willing to look closely enough.

References and Further Reading

Foundational Research

1. Bommasani, R., et al. (2021). “On the Opportunities and Risks of Foundation Models.” Stanford HAI Center for Research on Foundation Models. [Comprehensive analysis of bias in large language models]

2. Bender, E., et al. (2021). “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?” FAccT ‘21. [Seminal paper on data bias and environmental/social costs]

3. Perez, E., et al. (2022). “Discovering Language Model Behaviors with Model-Written Evaluations.” OpenAI Research. [Demonstrates prompt sensitivity and interaction bias]

4. Maynez, J., et al. (2020). “On Faithfulness and Factuality in Abstractive Summarization.” ACL 2020. [Critical for understanding hallucination patterns]

5. Welleck, S., et al. (2023). “Generating Sequences by Learning to Self-Correct.” ICLR 2023. [Evidence for adversarial prompting effectiveness]

Legal and Ethical Frameworks

6. American Bar Association (2024). “Resolution 604: Generative AI and the Legal Profession.” [Establishes duty of technological competence]

7. Legg, M., & Song, J. (2024). “AI in International Arbitration: Promise and Peril.” Journal of International Arbitration, 41(2). [Specific to arbitration context]

8. Whittlestone, J., et al. (2019). “The Role and Limits of Principles in AI Ethics.” Journal of AI Ethics. [Philosophical foundations]

Practical Guides

9. Anthropic (2024). “Constitutional AI: Harmlessness from AI Feedback.” Anthropic Research. [Understanding design bias in Claude models]

10. OpenAI (2023). “GPT-4 Technical Report.” OpenAI Research. [Transparency about model capabilities and limitations]

11. Mollick, E. (2024). “Co-Intelligence: Living and Working with AI.” Portfolio/Penguin. [Accessible introduction to practical AI use]

Specialized Resources for Arbitrators

12. Malek, B., et al. (2024). "SVAMC Guidelines on the Use of Artificial Intelligence in Arbitration." Silicon Valley Arbitration & Mediation Center. [First comprehensive guidelines for AI use in arbitration practice]

13. International Chamber of Commerce (2024). “Guidance Note on the Use of Artificial Intelligence in ICC Arbitration.” [Practice-specific guidance]

14. Global Arbitration Review (2024). “AI and Arbitration: A Practical Guide.” [Case studies and practitioner perspectives]

15. UNCITRAL (2023). “Technical Notes on Online Dispute Resolution.” [Framework applicable to AI-assisted processes]

Model-Specific Documentation

15. Claude Documentation: https://docs.anthropic.com

16. OpenAI Documentation: https://platform.openai.com/docs

17. Google AI Documentation: https://ai.google.dev

18. Meta Llama Documentation: https://llama.meta.com

Ongoing Resources

19. Stanford HAI: https://hai.stanford.edu [Regular updates on AI bias research]

20. Partnership on AI: https://partnershiponai.org [Cross-organizational guidance]

21. AI & Law Forum: Various university law schools maintain active research programs

Note to Readers: This article reflects the state of AI technology and arbitration practice as of January 2026. Both fields are evolving rapidly. For current information on specific models or practices, consult the documentation and research resources listed above.

This article is researched and illustrated with AI tools (ChatGPT, Claude and Midjourney)

This is part of TechCred’s ongoing series on practical AI literacy for legal professionals. For questions or to suggest future topics, reach out through Substack.

Arbitration at the Forefront of the AI Era

Generative AI is reshaping how we think about evidence. Synthetic media—whether video, audio, or documents—has reached a level of realism that raises important questions about authentication. In litigation, these developments are sometimes framed as a crisis. In arbitration, they can be viewed as an opportunity. Arbitration’s hallmark flexibility makes it uniquely well-suited to adapt quickly, refine safeguards, and build trust in a digital era.

A widely reported 2024 case in Hong Kong, where deepfake video conferencing tricked a multinational firm into transferring millions, illustrates the technology’s power. Yet it also highlights arbitration’s advantage: unlike courts constrained by legislation, arbitral tribunals can experiment with innovative tools and protocols, positioning arbitration as a leader in shaping global standards for evidence integrity.

Arbitration’s Adaptability as an Asset

Arbitration has always thrived by combining flexibility with fairness. Party autonomy under the UNCITRAL Model Law allows procedures to be tailored to the dispute, while institutions like the ICC and LCIA set broad but adaptable frameworks. That same flexibility can be used to integrate new safeguards for synthetic evidence.

While some see vulnerability, others see agility. Courts must wait for legislatures to update statutory rules. Arbitral tribunals, by contrast, can respond immediately—through case management orders, disclosure protocols, and expert appointments. A 2025 Queen Mary/White & Case survey confirmed that arbitrators already anticipate a growing role for AI in proceedings, which means readiness is top of mind across the field.

Authentication as a Constructive Challenge

In the depths of a dispute, where light is scarce and appearances can mislead, deepfakes raise two key questions: how to authenticate digital evidence, and how to preserve procedural fairness.

• Authentication: Tools for forensic verification are improving rapidly, from metadata analysis to AI-driven detection software. Tribunals can require these tools as a matter of procedure, turning uncertainty into a structured evidentiary process.

• Procedural fairness: Far from undermining due process, these challenges give tribunals the chance to refine practices like cross-examination, disclosure, and expert testimony. By doing so, arbitration can offer a model of how justice adapts responsibly to technology.

Rather than assuming risks will overwhelm proceedings, these questions become opportunities to demonstrate resilience and creativity.

Practical Pathways for Tribunals

Arbitration can stay ahead of synthetic evidence by integrating solutions that are already emerging:

1. Metadata and Verification
   Require parties to provide metadata with digital submissions and enable forensic screening. The IBA Rules on Evidence already allow tribunals to assess probative value and exclude unreliable materials.

2. Appointment of Neutral Experts
   ICC and UNCITRAL frameworks empower tribunals to appoint experts. Applying this to AI forensics ensures impartial analysis of contested evidence.

3. Disclosure Protocols
   Tribunals can request early disclosure of whether AI tools were used in preparing evidence. Transparency prevents disputes from being derailed by surprise allegations.

4. Immutable Record-Keeping
   Secure timestamping and blockchain-based chains of custody can reinforce confidence in digital submissions, offering parties reassurance that evidence cannot be altered mid-proceeding.

5. Institutional Guidance
   Initiatives within ICC, CIArb, and SVAMC are exploring how AI intersects with arbitration. Tribunals can draw on this momentum and adopt new recommendations proactively at the case-management stage.

Each of these measures is less about guarding against disaster and more about reinforcing the credibility of arbitration as forward-thinking and robust.

Conclusion: Versatility Through Adaptability

Deepfake and synthetic-evidence threats challenge arbitration’s resilience. In navigating these deep waters, arbitrators can chart the course, showing how fairness and trust endure even in the darkest depths.

Synthetic evidence is best seen not as a crisis but as a proving ground for arbitration’s adaptability. With proactive integration of technology and updated institutional guidance, arbitral tribunals can turn what some see as a threat into an advantage.

The lesson is not that arbitration is vulnerable, but that it is versatile. As AI reshapes the evidentiary landscape, arbitration can prove again why it remains the preferred forum for complex, cross-border disputes: flexible, trusted, and ready to adapt.

Further Reading

• Financial Times, “Hong Kong Engineers Fooled in $25m Deepfake Video Scam” (2024) – A real-world example of how synthetic media is reshaping trust in digital communications.

• Berkeley Technology Law Journal, “Synthetic Evidence in Litigation: Authentication and Adaptation” (2024) – Academic discussion on how courts and tribunals can adapt evidentiary practices.

• ICC, Task Force on the Use of Information Technology in International Arbitration – Ongoing institutional work exploring technology’s impact on arbitral procedure.

• CIArb, Guidelines on the Use of Technology in International Arbitration (2024) – Practical recommendations for tribunals managing digital evidence and AI-related issues.

• UNCITRAL, Notes on Online Dispute Resolution (2021) – Guidance on managing digital submissions and remote hearings.

• SVAMC, Commentary on AI in Arbitration (2024) – Early thought-leadership on disclosure and transparency when AI tools are used in evidence preparation.

• Queen Mary University of London & White & Case, 2025 International Arbitration Survey – Insights into how arbitrators and practitioners view AI’s role in current and future disputes.

TechCred Session II (Oct 17, 2025) :
AI, Autonomy & Evidence That Doesn’t Sleep

These same themes echoed across Session II of TechCred’s inaugural program, where leading voices in the field examined the evolving responsibilities of arbitrators in the face of synthetic evidence, AI-assisted reasoning, and decentralized legal actors.

The message was clear: arbitration is no longer reacting to technological change—it is defining the standards by which justice will operate in the digital era.

////

This article is written together with AIs.

In his latest article, Barry Appleton argues that Canada must confront a growing crisis of algorithmic governance. Our digital reality—what news reaches us, what opinions circulate, what voices are amplified—is increasingly determined by a handful of global platforms whose operations remain opaque, unaccountable, and largely unregulated.

Appleton points out that this is not just a matter of consumer protection or data privacy. It is a constitutional issue. When platforms act as arbiters of speech and access to information, they effectively perform a public function—without public oversight.

But what happens when a country’s legal framework is built for human decision-makers, and decisions are now being made by code?

Courts can subpoena a minister, cross-examine a witness, or compel disclosure—but they cannot cross-examine an algorithm. Yet algorithms shape perception itself. They decide which facts gain traction, which opinions appear balanced, and which truths are quietly buried beneath an ocean of noise.

True oversight, then, requires more than transparency; it demands comprehensibility. We must be able to understand not just what algorithms are doing, but why—and to ensure that their reasoning aligns with our democratic values.

As Appleton writes,

“Algorithmic systems have replaced the editorial judgments of humans, but not their responsibilities.”

The challenge is to bring law, technology, and ethics back into conversation—to make sure those responsibilities are not lost in translation between code and conscience.

The TechCred accreditation program exists for precisely this reason: to prepare lawyers, arbitrators, and policymakers for a world where code itself can be a decision-maker. Algorithmic oversight is not only a regulatory challenge—it is a matter of justice, transparency, and democracy.

Further Reading

Barry Appleton

Read Barry’s full article: Algorithmic Oversight: Canada Must Confront Platforms That Control What We See
(Originally published in National Post, October 7, 2025)

.

Subscribe now

Written together with AIs

When you launch something that’s never been done before, the support of strong partners makes all the difference. We’re deeply grateful to our co-sponsoring organizations—each helping us build something so critical at this point in time. To reflect our appreciation, every partner received a 40% discount code for their members and associates. We’re thrilled to have such formidable allies alongside us as we launch the inaugural TechCred program.

Visit their social media channels to find their code.

If you had to describe the digital age in three words, you could do worse than these:
Fake identities. Fake information. Fake behaviour.

That’s how a recent paper by TechCred faculty member Dmytro Chumachenko and colleagues frames some of the new challenges created by generative AI.
Not threats in the sci-fi sense, but shifts — quiet, steady shifts — in how we recognize what’s genuine in a world increasingly built on code.

A fake identity might be a composite photo of a person who doesn’t exist.
Fake information might be text written by a machine that sounds perfectly human.
Fake behaviour might be a swarm of invisible bots giving the illusion of a public reaction that never happened.

If that sounds abstract, think of how often we already rely on unseen systems to make decisions — search engines, credit scoring, recommendations, even legal analysis. These systems don’t always deceive us, but they do mediate reality. The “three fakes” remind us that what we see and what we trust are now designed experiences. Recognizing that isn’t cynicism; it’s the beginning of digital literacy for an algorithmic world.

Rather than seeing this as a crisis, we might see it as an invitation:
to rebuild trust with intention, to design systems that value transparency, and to remember that judgment — real human judgment — is still the most advanced technology we have.

Maybe that’s the work ahead.
And maybe that’s what TechCred has always been about.

REGISTER NOW

Read the full Balsillie Paper: Artificial Intelligence and New Threat Vectors: Using Scenario Planning for Trend Forecasting

.

This piece was written together with AIs.

.

Ed Mullins is a Partner at Reed Smith in Miami, board-certified by the Florida Bar for International Litigation & Arbitration. His four-decade career encompasses high-stakes investor-state arbitration, energy and construction disputes, distribution and joint venture litigation, and class-action defense. He also serves as arbitrator and is a member of the AAA Commercial Panel and a fellow of the Chartered Institute of Arbitrators

An accomplished appellate litigator, Ed has argued over 100 appeals—including before the U.S. Supreme Court—and played a pivotal role in establishing Florida's board certification in international arbitration . He is Chair-Elect of the ABA Section of International Law and serves in multiple leadership capacities within the ABA and Florida Bar, including co-leading arbitration and ADR committees. His peers consistently recognize him, with distinction from the Best Lawyers, Legal 500, Chambers USA (Band 2), and Miami Legal Elite listings

References

• Ed Mullins – Reed Smith Professional Profile & Practice Focus

• Chambers USA – Litigation: General Commercial (Florida), Band 2

• Florida Bar Board Certification & Profile

• Reed Smith News: Ed Mullins Reappointed as ABA Revenue Officer

By Artem Artyukhov (Slovakia, Ukraine), Dmytro Chumachenko (USA, Ukraine) published in Geopolitics under Globalization, Volume 6, 2025

Abstract

This study aims to examine the role of international organizations in promoting academic integrity, analyzing how diverse cultural interpretations of scholarly ethics create challenges for standard consistency across borders and the mechanisms through which international bodies facilitate the coordination and standardization of integrity practices. A comparative approach examines academic integrity across different jurisdictions and cultural contexts, a multi-level analytical framework understands the organizational structure of academic integrity governance, and illustrative case studies demonstrate practical applications. The analysis identifies five interconnected components of academic integrity: educational, research, managerial, professional association, and publishing integrity.

This study reveals a sophisticated five-level hierarchical system of academic integrity governance spanning from international organizations to individual practitioners. International organizations function as collaborative facilitators rather than rigid rule-makers, developing flexible frameworks that can be adapted to diverse cultural contexts while maintaining universal ethical principles. Significant variations exist between national approaches, with some countries implementing comprehensive legislative frameworks while others relying on cultural principles and institutional traditions. International organizations facilitate dialogue and consensus-building that transcends national boundaries while respecting institutional autonomy.

The hierarchical governance system demonstrates that academic integrity assurance requires top-down coordination to establish consistent standards and bottom-up commitment to implement those standards. The strength of this system lies in its ability to maintain the universality of core ethical principles while allowing for cultural adaptation in implementation methods, ultimately ensuring that the fundamental commitment to honesty, originality, and fairness in scholarship remains constant regardless of cultural context or geographical location.

FULL PAPER

TechCred notes:

• Relevance: This study aligns with TechCred’s focus on governance in the algorithmic era. It addresses how international organizations craft flexible, culturally adaptive frameworks—paralleling the challenge of developing AI governance models that maintain ethical consistency while accommodating jurisdictional diversity. Its insights into multilevel coordination and institutional legitimacy resonate with the regulatory questions facing arbitral institutions as they integrate digital tools and standards.

• Cross-disciplinary impact: Bridging education policy, ethics, and international governance, the paper offers a valuable analog for AI regulation and digital integrity systems. It provides conceptual scaffolding for understanding how norms of transparency, accountability, and fairness can be operationalized across borders—making it relevant not only to arbitrators and policymakers but also to technologists and academic leaders shaping digital-era ethics frameworks.

Governance once meant parliaments, ministries, and courts. Today, it’s also written in code — in how data moves, how models learn, and how rules are automated. Each time a policymaker uses an AI tool or a tribunal evaluates digital evidence, a question surfaces: who governs the governor?

Law now lives inside architecture. Every algorithm is a policy choice, every dataset a jurisdictional line. As systems cross borders, their logic does too. Institutions that once defined ethical norms—universities, courts, regulators—now depend on technologies governed elsewhere. This isn’t conspiracy; it’s inertia. We built global infrastructure before we built global legitimacy.

Dr. Dmytro Chumachenko’s research on academic integrity offers a model for this moment. He maps five layers of governance—international, national, institutional, professional, and individual—showing how ethics can survive across borders through coordination, not control.

Replace academic integrity with algorithmic integrity and the principle still stands: legitimacy in complex systems is not imposed, it’s negotiated.

The problem is that the world’s digital scaffolding remains largely private and opaque. We regulate what we see—the apps, the outputs—but not the layers beneath. Cloud contracts, inference engines, and content filters now act as meta-laws, shaping what can be known or believed. Without oversight, automation risks becoming the illusion of governance. Speed isn’t legitimacy.

Barry Appleton calls this the new form of sovereignty—rule by computation. Chumachenko reminds us that integrity is never held by one level alone. It depends on constant dialogue between global principles and local realities. True digital sovereignty is not the power to exclude, but the ability to participate responsibly in systems that include everyone.

If the nineteenth century asked who owns the land, and the twentieth who owns the oil, the twenty-first asks who owns the logic. The answer will determine whose values are encoded into the systems that govern daily life. Governance must evolve from regulation to architecture—from writing rules to designing systems that make fairness the default.

That is the new test of legitimacy.

This article was inspired by recently published works by Prof. Barry Appleton and Dr. Dmytro Chumachenko

Read more

Barry Appleton: AI and cloud infrastructure is the railway of the future — why isn’t Canada building it?

Published September 30, 2025 National Post

Artem Artyukhov & Dmytro Chumachenko: International organizations in promoting academic integrity in a globalized world: Ukrainian view

Published July 9, 2025 Geopolitics under Globalization Journal Volume 6

/// Written together with AIs.

The TechCred Faculty is an elite group of globally recognized experts at the intersection of international arbitration, law, and emerging technology. Each faculty member brings a distinguished track record in cross-border dispute resolution, innovation policy, and legal technology, making this one of the most qualified bodies assembled to address the challenges and opportunities of technology in international arbitration.

The school is a hub for thought leadership and innovation in international law, dispute resolution, and digital economy policy. The New York Law School is pleased to support the American Bar Association’s TechCred program’s curriculum and credentialing. TechCred is supported through the New York Law School’s Center for International Law.

TechCred Faculty Members

• Prof. Barry Appleton – Director, TechCred Faculty Co-Director, Center for International Law, New York Law School & Fellow at Balsillie School of International Affairs, International Arbitration Counsel. A pre-eminent practitioner and scholar on international arbitration, global business, and regulation of the digital economy, Prof. Appleton has acted in landmark disputes and authored leading publications on NAFTA, investment arbitration, and multilateral trade law. His practice focuses on the impact of the emerging digital economy on international arbitration and international economic law.

• Dr. Dmytro Chumachenko, Associate Professor, Department of Mathematical Modelling and Artificial Intelligence of National Aerospace University “Kharkiv Aviation Institute” (Ukraine), Scholar, Balsillie School of International Affairs. Research Associate at the Massachusetts Institute of Technology (USA), Research Affiliate at the University of Waterloo (Canada), Visiting Researcher at the Max Planck Institute for Demographic Research (Germany), and Invited Lecturer at the Wildau Technical University of Applied Sciences (Germany).A scholar specializing in cybersecurity, AI governance, and international institutions, Dr. Chumachenko provides thought leadership on the regulatory frameworks needed for fair and secure arbitration in a digital age. He is Chair of the Education and Science subcommittee of the Expert Committee on Artificial Intelligence Development under the Ministry of Digital Transformation of Ukraine and Chair of the Board of the Ukrainian Scientific and Educational IT Society.

• Daniel Garrie, Neutral, JAMS; Executive Managing Partner, Law & Forensics.

  Daniel Garrie is a JAMS Neutral, Executive Managing Partner at Law & Forensics, and adjunct faculty at Harvard. An experienced international arbitrator, he is recognized for resolving complex disputes involving artificial intelligence, digital forensics, cybersecurity, and e-discovery. Combining deep legal and technical expertise, Mr. Garrie has worked with courts and arbitral tribunals, bringing clarity to cross-border disputes at the forefront of the digital economy.

• Prof. Maura R. Grossman - University of Waterloo and Osgoode Hall Law School - A senior computer scientist and law professor, Prof. Grossman is a groundbreaking expert in technology-assisted review (TAR) and e-discovery. Professor Grossman’s scholarship is internationally cited in leading arbitral and judicial decisions regarding legal technology and digital evidence. Her work is considered the gold standard on evaluating e-evidence.

• Annie Lespérance, Head of North America for Jus Mundi - Ms. Lespérance is an arbitrator and a rising figure in technology-informed dispute resolution. She is recognized for her efforts in promoting digital literacy in arbitration and for bridging the common and civil law approaches to tech disputes.

• Dana MacGrath, Arbitrator, MacGrath Arbitration, past president, ArbitralWomen, Ms. MacGrath is deeply engaged in thought leadership around the modernization of arbitral practice. She frequently lectures on environmentally friendly arbitration procedures and the integration of technology in arbitrations. A highly respected international arbitrator with over two decades of experience, she is a recognized leader in the field of international commercial arbitration, with expertise in complex cross-border disputes involving energy, finance, and emerging technologies. Her practical insights, academic rigor, and advocacy for innovation make her a vital member of the TechCred Faculty.

• Benjamin Malek, T.H.E. Chambers, Founder, LexArb.Board Member and Former Chair of the Silicon Valley Arbitration and Mediation Center (SVAMC) AI Task Force. Mr. Malek initiated the SVAMC AI task force, which led to the development of one of the world’s first Guidelines on AI in Arbitration. He continues to demonstrate this thought leadership at LexArb, a tech company specializing in AI-driven arbitration solutions. Mr. Malek focuses on digital evidence, digital asset litigation, smart contracts, and blockchain-based arbitration disputes, making him a key figure in cutting-edge commercial arbitration involving decentralized technologies.

• Rebeca Mosquera, Senior Counsel, Saudi Aramco, President, ArbitralWomen. Rebeca Mosquera is a global leader in international arbitration, with a practice spanning Latin America, Europe, and the United States. Deeply committed to AI in arbitration. As the President of Global Women, and with her distinguished disputes practice, she brings a strong focus on digital innovation, inclusion, ethics, and modern arbitral practice.

• Kathleen Paisley, Principal, AMBOS Lawyers (Brussels/Miami), -Arbitrator and Digital Law authority. One of the rare arbitration practitioners dually qualified in law and finance, Ms. Paisley brings unmatched expertise in data protection, cross-border finance, and international commercial arbitration, with particular insight into GDPR and tech-regulated markets.

• Sarah Reynolds, Arbitrator and Board Member, Silicon Valley Arbitration & Mediation Center (SVAMC). Ms. Reynolds is one of the leading arbitration counsel in the technology sector, with a focus on high-stakes commercial disputes involving digital platforms, algorithms, and AI-related claims.

A Faculty of Global Leaders in Tech-Driven Arbitration

Collectively, the TechCred Faculty represents:

• Leadership roles in arbitration bodies

• Faculty appointments at top global institutions, including New York Law School, Harvard, MIT, University of Waterloo, Osgoode Hall, Max Planck Institute for Demographic Research, and Balsillie School of International Affairs

• Direct experience with ICC, JAMS, ICDR, VIAC, UNCITRAL, ICSID, SVAMC, and ad hoc arbitral tribunals

• Pioneering expertise in AI, e-discovery, digital forensics, quantum computing, blockchain, and cybersecurity law

This faculty’s combined experience and credentials position TechCred at the forefront of shaping how international arbitration adapts to, and leads, the digital transformation of legal practice.

REGISTER NOW

///

William K. Hill is a shareholder at Gunster, based in Miami, with over 30 years of experience as a business advocate, litigator, and arbitrator. He is the only attorney in Florida board-certified in both Business Litigation and International Litigation & Arbitration. William represents sectors including technology, healthcare, aviation, real estate, and finance, handling complex commercial matters, intellectual property disputes, and international arbitrations. His peers recognized him in 2025 as the AAA/ICDR Rising Star Arbitrator, a distinction based on consistently high client and counsel ratings.

A seasoned ADR professional, William regularly serves as arbitrator and mediator, appearing on the Roster of Neutrals for AAA and ICDR, and holding Fellow status with the Chartered Institute of Arbitrators. He is a founding board member of the Miami International Arbitration Society (MIAS), and actively involved in the ABA, including as a Revenue Officer in the International Law Section. William is also noted for engaging in regulatory developments, such as commenting on recent amendments to Florida’s court-appointed arbitrator rules.

References

• Gunster News: AAA Rising Star Arbitrator 2025

• Gunster Professional Profile – William K. Hill

• Miami International Arbitration Society – Board of Directors

• Florida Bar – Board Certification in Business Litigation and International Litigation & Arbitration

Benjamin I. Malek, LL.M., FCIArb is Partner and Arbitrator at T.H.E. Chambers in New York, with over a decade of experience in international arbitration. He has handled commercial, investor-state, and state-to-state disputes under ICC, ICDR, AAA, LCIA, SIAC, HKIAC, CPR, IACAC, THAC, ICSID, UNCITRAL, and ad hoc rules. His work spans multiple legal traditions across the U.S., Latin America, Central & Eastern Europe, and the Middle East, giving him fluency in civil- and common-law contexts.

Benjamin is Chair of the Silicon Valley Arbitration & Mediation Center (SVAMC) AI Task Force and the principal author of the SVAMC Guidelines on the Use of Artificial Intelligence in Arbitration—a soft-law framework widely respected by institutions and practitioners. He is multilingual in English (native), Spanish, French, German, Romanian (professional proficiency), with some Italian and Mandarin. His broad sectoral scope includes space, web3, cybersecurity, biotech, energy, infrastructure, and more.

His leadership at the nexus of technology, institutional evolution, and global arbitration speaks to a future where excellence in ADR is defined not just by tradition, but by forward-leaning strategy and cross-disciplinary fluency.

References

• Benjamin I. Malek – T.H.E. Chambers Profile

• SVAMC AI Task Force Chair – Reed Smith Newsletter (Note: referenced via Reed Smith commentary on SVAMC AI Guidelines)

• LinkedIn Profile

Sarah Reynolds, FCIArb, is a respected arbitrator and counsel with deep experience representing technology and pharmaceutical companies in both domestic and international arbitrations. Licensed in Illinois, California, England & Wales, and Ireland, she has served at leading firms such as Mayer Brown and Goldman Ismail Tomaselli Brennan & Baum LLP, and now practices independently via Reynolds ADR. She sits on key panels—including ICDR, the British Virgin Islands Arbitration Center, and the Hong Kong International Arbitration Centre—bringing cross-jurisdictional insight to her cases.

A former CEO of the Silicon Valley Arbitration & Mediation Center (SVAMC) and now a Board and Executive Committee Member, Sarah continues to drive innovation in arbitration’s intersection with technology. She is regularly invited to expert panels—such as CIArb-NAB’s “Tech ADR: Insights from the Silicon Valley Vanguard”—to discuss evolving ADR topics like deepfakes, data privacy, and AI-assisted proceedings.

References

• SVAMC — Member Profile: Sarah Reynolds

• CIArb Member Directory: Sarah E. Reynolds

• CIArb-NAB Event: "Tech ADR: Insights from the Silicon Valley Vanguard" featuring Sarah Reynolds