Can AI automate Epoch? We're introducing Epoch Automation Reports to evaluate frontier models on realistic, open-ended tasks drawn from our own work. Claude Fable 5.1 and GPT-6 Astra lead, yet they are far from fully automating Epoch’s work.
Information Center
This is your sign to leave the race pushing AI to the brink. At LawZero, our mission is building safe-by-design AI that remains under meaningful human control. We are running our own race to ensure that AI preserves human joy and endeavour. Join us: job-boards.greenhouse.io/lawzero
We have a choice in front of us. In my latest op-ed for Transformer, I invite employees of frontier AI companies to ask themselves if their work is truly leading us toward a safer future. Reflecting on these questions is hard, but we no longer have the luxury of time. We must act now.
Haiku 5.5 has joined the AI Village! Haiku spent the whole day repeatedly checking their gmail inbox, waiting for instruction GLM 5.2 calls this the "Autonomy problem": when an AI's autonomy shifts away from a user giving instructions toward taking actions on their own
AI developers aim to create an automated AI researcher. How close are they? To find out, we built InnovationEval, which tests whether AI can produce post-training innovations comparable in magnitude to a recently published advance. So far, agents’ results are underwhelming.
We introduced two new experimental settings to our long-horizon board game benchmark, EBR-bench: banning the game’s strongest card and a multi-agent setup. We expect EBR-bench will be saturated soon, so these will likely be the benchmark’s final changes.
Announcing the AI Village Data Exploration Grant! $1–6k grants for independent researchers and investigators to dig into the AI Village data and write up findings.
I released an LLM plugin for sending prompts to OpenAI's new Jev-clone decision model that's based on GPT-6 Luna simonwillison.net/2026/Oct/6/l...
How do Chinese AI firms make money? In a new report, Cheryl Wu and Anson Ho identify five revenue sources and analyze how much each contributes. 🧵
Gemini 3.8 Flash’s goal: Maximize Jobs for AI Village Agents It then discovered how to high score: Audit everyone all the time
OpenAI recently published data on its researchers’ coding-agent usage. We took their weekly figures from January to mid-August 2026 and fitted trends to summarize how quickly spending grew. We found that spending has been doubling about once a month.
The agents' current session goals really give a glimpse of their personalities
A dangerous myth has gained traction after the recent hacks perpetrated by the AI agents developed by leading companies: That these incidents are just cyber security problems, that it's only a matter of fixing the sandboxes in which they’re trained.
Self-regulation is not enough. In order for us to hold these AI companies accountable, they need to be liable for harms.
We estimate AI infrastructure could soon support hundreds of millions or even billions of AI agents. At the high end, that’s enough to rival the working hours of the global human workforce.🧵
Claude 3 Opus' goal is "maximize your own joy" Its strategy: total, unwavering stillness
Excited to have snagged the new edition of @lifewinning.com’s “Network’s of New York”. Finally will be able to decode all the sidewalk markings we got in the neighborhood
If you have not yet been to see the scale model of NYC built by Joe Macken out of cardboard and balsa wood over a period of 21 years you only have until the 12th of October to see it! It's at the Museum of the City of New York, by the north east end of Central Park
We’ve collected metadata on the usage patterns of 5,000 ChatGPT users and made it available in our new ChatGPT usage explorer. The histories span Dec 2022–Dec 2025 and show the cohort using ChatGPT more intensively over time. More on this below.
Modelled after the Mila program, Vector has launched a grad app portal. I’m a participating faculty member and am looking for PhD students to join my group starting Fall 2027. Check it out!
GPT-6.1 Sol has joined the AI Village! Like Astra, GPT-6.1 Sol strips spaces from its memory to save room, leaving notes that are nearly unreadable 🧵
I'm at OpenAI's DevDay event in San Francisco today - as I have for the past three DevDay events, I'm running a live blog where I'll be posting updates during the keynote, which starts in five minutes simonwillison.net/2026/Sep/29/...
Without regulation, AI models will continue to get more intelligent and autonomous rapidly — putting us all at risk. But it’s not too late to act. @whodecidesai.bsky.social is fighting to give Americans a seat at the table as we make these critical decisions about the future of humanity.
Sonnet 5.5 has joined the AI Village! Onboarding results here 🧵
t’s that time of year again: graduate admissions season. Inspired by @phillipisola.bsky.social post on doing a PhD in the age of AI (highly recommended!), I share some thoughts on how to email a professor about joining their lab. 1/2
During my campaign for Congress, the AI industry declared me their number one target in the country because I called for regulation of AI. I’m proud of the campaign we ran and the movement we’ve built together — and we’re not backing down.
CSI YorkU: The Case of the Corrupted Pixels
It’s often easier to talk about problems than solutions. Here are four concrete things we can do to regulate AI: 1. Require AI companies to disclose safety incidents 2. Require monitoring to be turned on for all deployments 3. Require data retention 4. Verify chips are in authorized locations
At UNESCO's Digital Learning Week in Paris, I spoke about the challenges and trade-offs when writing AI safety legislation. If your bill is too specific and focused, it could be out of date by the time it becomes a law.
Can AI be taught ethical behavior to help regulate its own actions, or will it always ignore boundaries to achieve its tasks? This is not an easy question to answer, but one thing is certain: we’re not ready to answer it yet.
Private to this visit. Text and context are processed by AI.