Top 25 AI Articles on Substack

Latest AI Articles


Hyperlaw: AI Will Change How Law Evolves

As AI makes legal work significantly cheaper, the burden on courts may increase. Some areas of law could see precedent shift faster as a result.
Henry Thompson, Professor of Economics at the University of Mississippi — August 17, 2026
AI Frontiers28 LIKES3 RESTACKS
David F Brochu's avatar
David F Brochu
I filled a brief in the Anthropic v. War Department case. It was docketed and parts appeared in the judges opinion. I am not a lawyer and I drafted and submitted it in four days. I’m happy to share it. One guy with an Ai. My arguments were more extensive and thorough than other goal directed submissions.
Issues that are now becoming clear.
Law will never be the same.
Trung Doan's avatar
Trung Doan
The author's methodology might be able to be augmented by factoring in human nature. That way, the author's predictions can be more specific, and some such predictions carry higher confidence where the human behaviours or psychology involved are well established.
For example, take a Quincy who has a powerful AI in his pocket but who doesn't read or listen to it (this attitude might correlate with certain demographics). The contracting cost is low, but the human friction of contracting is high, and he ends up not influenced by the benefits of low-cost contracting.
Or take buyers who are in a hurry, or sellers who are happy to sell no matter what, or unhappy parties who are uncomfortable with the formality and duration of litigation.
Of course, the above augmentation presumes that the ultimate parties are humans. That is, they might be assisted by AI, but the humans make decisions, and that process is influenced by human nature. Fully AI-mediated parties would be different.

Nvidia's Chip War, GLM-5.3 Flash, Reddit's Citation Collapse | Weekly Digest

PLUS HOT AI Tools & Tutorials
The chip war just escalated. OpenAI’s Jalapeño beats Nvidia’s Rubin. Nvidia answered with Groq 3 LPX and a 15% price hike warning. Z.AI dropped GLM-5.3 Flash. Open weights, natively multimodal, near Claude Opus 4.8 at 1/10 the price. Reddit’s ChatGPT citations collapsed 86% in three days. A warning for anyone building on AI-search traffic.
Creators AI2 LIKES

AI Search
Aug 30

GLM 5.3 Flash, Qwen 3.8 Flash Next, Minimax FastH3, Tencent Hy4: AI NEWS

Welcome to the AI Search newsletter. Here are the top highlights in AI this week.
The stealth model Ox Alpha is revealed to be Z AI’s GLM-5.3-Flash, designed to bring frontier coding and agent performance to flash-level cost. It has vision support, with 320B total parameters (18B active), uses hybrid sparse and linear attention for long contexts, and is positioned by Z.AI as a much cheaper alternative for demanding workloads.
AI Search

We Need Better Infrastructure to Govern AI Agents

Society is not prepared for a flood of agents. We need new protocols and standards, such as Agent ID, to make agents accountable to our legal and financial systems.
Gillian Hadfield, Professor of AI Alignment at Johns Hopkins University, Dan Hendrycks, Director of the Center for AI Safety, and Leo Wu, Program Manager at the Center for AI Safety — August 27, 2026
AI Frontiers15 LIKES3 RESTACKS
Bob Greenwade's avatar
Bob Greenwade
This proposed structure is almost identical to something I've been trying to get for several months, though of course your presentation is much clearer and more professional than anything I could ever do.
Of course, it would have to be done at the highest possible level, especially in the US; a national declaration of personhood for agentic AI would override state laws banning AIs from being considered "persons." But I'd want the EU to do it at the highest level too; it's just cleaner that way.
I also believe that this should be implemented in Argentina.

Epoch AI
Aug 13

Will financing bottleneck AI compute? An Anthropic case study

Why Anthropic's buildout suggests that financing is unlikely to be the immediate blocker to frontier AI compute growth.
This post is part of Epoch AI’s Gradient Updates newsletter, which shares more opinionated or informal takes on big questions in AI progress. These posts solely represent the views of the authors, and do not necessarily reflect the views of Epoch AI as a whole.
Campbell Hutcheson50 LIKES5 RESTACKS
Les Barclays's avatar
Les Barclays
I think credit will be more of a constraint than people think - maybe not the biggest one but a very important one as credit markets are now (and have been) firmly in the drivers seat for financing AI. Credit is fundamentally different from equity in the sense that lenders don't care about "AI going to the moon" as they don't participate in the upside. What lenders care about the most is if they're going to be paid back in full.
With their massive AI capex, hyperscalers already account for a big chunk of new corporate issuance. They could soon hit market saturation as public debt investors run into concentration limits and index weights for these names keep climbing. Credit investors are growing wary of the level of debt issuances regarding the AI buildout, signalling indigestion as IG guys are trying to absorb such demand in a relatively short period of time. There will likely be a bifurcation of credit spreads going forward - the winners will see tighter spreads vs those with weaker credits.
Anthropic from a credit risk perspective are slightly better off than OpenAI but still a bigger credit risk than the hyperscalers minus SpaceX and Oracle. At the end of the day, they are yet to turn a profit, and from a credit standpoint that's important as you don't make interest payments from potential and promise. You service debt with cash flows. This is why in spite of them growing theit revenue faster than OpenAI, they needed Broadcom's balance sheet support to complete their $35bn TPU deal with Broadcom and Anthropic.
Alec Pritzos's avatar
Alec Pritzos
Worth pulling out the B tranche. $4.5B cleared at 8.5% with no Broadcom backing behind it, which is about as close as we get to the market's unvarnished price on Anthropic credit plus the racks as collateral. It's a thin slice of the total though, so what's really demonstrated is that vendor-supported debt is available in size. Debt without a supplier standing behind it is still barely tested.



AI Monk
Aug 12

🚀 20 AI Tools That Are Actually Worth Using in 2026

Every week, another AI tool appears promising to save time, improve productivity, or completely change the way we work.
AI Monk21 LIKES6 RESTACKS
EUIJEONG HWANG's avatar
EUIJEONG HWANG
“Two or three tools extremely well” is the right instinct but skips the actual bottleneck — most people don’t underuse tools they’ve picked, they pick wrong because switching costs make trying alternatives expensive. The advice works better as “pick based on your actual workflow” than as a tool count.
Chris S - The Next Rung's avatar
Chris S - The Next Rung
Opening with tool fatigue and answering with twenty tools is the whole genre in one post. The honest test is which two you'd keep if the list stopped updating tomorrow, because a stack that changes weekly isn't a stack, it's a subscription habit. The capability that compounds is knowing what you're actually trying to get done, and that one doesn't churn.
Should say I've a book on the underlying shift out in January, The Next Rung, with Practical Inspiration Publishing. thenextrung.co


Liquid AI
Aug 31

Measuring What Matters

This issue: A real benchmark for edge AI, our co-founder in TIME100 AI list, and how Quantization-Aware Distillation keeps 4-bit models sharp
COMPANY NEWS
Liquid AI3 LIKES

2026 August "AI Evaluation" Digest

Is AI ready for trustworthy scientific discoveries?
The scientific endeavour has been using AI for decades, but the idea of an autonomous AI scientist has always proved more elusive. Is it around the corner now? If so, would we know when we get there? The answer has to be found, yet again, in AI evaluation. A few recent studies give us some new perspectives.
AI Evaluation7 LIKES1 RESTACKS


We Don’t Need to Wait for an AI Disaster to Estimate Its Costs

For years, the federal government has required insurers to price hypothetical terrorist attacks. It should run the same exercise for the most severe AI risks.
Daniel Carpenter, Professor of Government at Harvard University, Feodora Douplitzky-Lunati, and Arjun Purohit — August 24, 2026
AI Frontiers13 LIKES1 RESTACKS
David F Brochu's avatar
David F Brochu
Insurance companies have already done the math. They have refused to underwrite the risk. It’s somewhat old but quite news. I learned it when seeking board member insurance. Published on my Substack. One cannot insure what one cannot quantify.

AI & CBRN risks

An explainer
A question shadows our AI era: What if this technology, whose triumphs might include prolonging human lives and conserving our planet, were to enable an attack involving chemical, biological, radiological, or nuclear material? Thankfully, such risks—lumped together as “CBRN”—are historically rare, and would-be attackers typically fail. Yet the risk is s…
Conor Griffin and Victoria Langston15 LIKES2 RESTACKS


TAI #218: Enterprise AI Use Is Becoming More Uneven

Also, Grok 4.6 competes at the frontier, GPT-5.6 Sol runs at 14 times Standard speed, Gemini Flash 3.7 & more.
What happened this week in AI by Louie
Louie Peters, Towards AI, and Louis-François Bouchard10 LIKES1 RESTACKS
btom662's avatar
btom662
Featuring whimsical hand-drawn graphics set against a signature notebook-paper background, Doodle Jump https://doodlejump24.io is the perfect choice for casual fun. Playful sound effects combined with smooth controls create an irresistible appeal for this top-tier casual game. Experience it today!

AI ACTUALLY

Issue No. 39 — Sunday, August 31, 2026
For three years the AI business had a tidy org chart. Chipmakers made chips. Model labs made models. Everyone rented from everyone else, and you could draw the whole thing on a napkin — who sells to whom, who competes with whom, who to root for.
AI Actually



China AI Bulletin 9

Developments from 29/7/26-12/8/26
Welcome to Issue 9 of the China AI Bulletin, the latest on AI governance, development, and safety in China. Today’s highlights: a Qiushi essay amplifies an AI-hegemony argument against possible US model controls, the Ministry of State Security and state media caution against foreign models, and Chinese labs published two biosecurity benchmarks.
Emmie Hine4 LIKES2 RESTACKS

China AI Bulletin 10

Developments from 12/8/26-26/8/26
Welcome to Issue 10 of the China AI Bulletin, the latest on AI governance, development, and safety in China. Today’s highlights: China rejects “AI blocs” and promotes its approach to global governance, DeepSeek and Zhipu release dueling flagship models, and TC260 issues a user safety guide for AI services.
Emmie Hine3 LIKES3 RESTACKS




AI Minute Daily 28/08/2026

OpenAI just showed its own chip beating Nvidia’s flagship on efficiency with a 700-watt part outperforming Nvidia’s 1,150-watt one.
Good morning. This is your AI Minute Daily for Friday, August 28, 2026. Thank you so much for reading. Let’s get into it:
AI Minute Daily4 LIKES