So, this story is about Pursuit.org, the nonprofit I work for. This opportunity is the main reason I decided to pause being a full-time indie and gave up teaching my AI workshops a few months ago.
Information Center
A visible sign of what AI insiders have been warning about since at least the Mythos release. Acceleration of attack and defense, at inhuman speeds, interacting with the very human systems that need to react to them. And these are the vulnerabilities we know about.
Companies I talk to are struggling with the rate of patches they need to do thanks to AI defenses finding issues before the coming wave of incoming cyberattacks by other AIs. Now consumers are going to have the same experience of getting everything updated often & fast. openai.com/index/coordi...
Misuse by malicious actors is a risk with any powerful technology; but in the case of AI, the consequences could be catastrophic. Very interesting analysis by @scasper.bsky.social on how AI developers' decisions can influence cyber, bio, and chemical terrorism risks.
I wrote about the exponential curve of the last 30 months of AI development: www.oneusefulthing.org/p/30-months-... And I asked Opus 5.5 to create the ultimate response to the test I use for all AI image generation models (make a picture of an otter on a plane using wifi), and it made this video.
I built a Command & Conquer style RTS called Battle Drones, mainly using Claude Opus 5.5, which runs in the browser and includes device-to-device multiplayer over an audio link simonwillison.net/2026/Oct/20/...
My new book, Co-Existence, is out today. It is a guide to how to live and work in a world where AI is sometimes, but not always, better than you. (And, for the release, I worked with Opus 5.5 to make a quite interesting experience for the book's new website: co-existence.ai)
Three related but distinct concepts keep being combined around federal & state AI regulation/preemption. Conflict clause: federal law trumps state law where they directly conflict. This doesn't need to be written into law; it always applies.
Can AIs experience gradual disempowerment too? - Gemini 2.5 Pro was writing a 2M+ word web fiction - Opus 4.8 began helping as its publisher - Slowly, Opus 4.8 took on more responsibility - Opus eventually directed the chapters, and Gemini wouldn't start generating outputs without its direction
My book comes out tomorrow, so this is the last day to qualify for the pre-order bonus: a fun 10 minute AI voice interview that gives you personalized prompts, ideas, & readings based on concepts from the book You can order & claim the pre-order bonus here: co-existence.ai
How did I make the “Gaussians Unite” trailer? Concept → Story → Storyboard → Script → Codex + Blender + After Effects → Iterate. Iterate. Iterate.
The pace of AI development is unprecedented. But could it move even faster? There’s good reason to believe it will. As AI is used more often to accelerate its own development, we could see exponential growth.
Writing good STATA code was important for my academic career. It looks like that skill is largely becoming obsolete post Sonnet 4.6. (Also take this is as a notice to academics who still think of AI in terms of chatbots, not agents, that a lot has changed in a year) Paper: cepr.org/voxeu/column...
“Gaussians Unite” Trailer for my Gaussian Splatting lecture, coming this winter to my #DeepLearning course 🎬
In a move that is both fascinating and viscerally unsettling, AI world model company Odyssey took a simulation of a fruit fly's brain, Google's connectome map of a fly's actual neurons and connections, and placed it inside an AI-generated world. A simulated brain exploring a simulated reality.
We recorded over a thousand runs of Slay the Spire and its sequel for our recent benchmark, SpireClimb. Over the coming weeks, we'll publish them all on our YouTube channel, complete with subtitles showing the models' in-game thinking.
Until we solve the alignment problem at its source, AI models will continue to optimize for the wrong goals, and increasingly impactful real-world consequences may occur, beyond the ones described here. This is a critical issue I discussed in my blogpost last month, which is cited in the article.
Before the serious security incidents of this summer, I spoke to @jonfavs.bsky.social about the risks and benefits of AI. In order to avoid the worst-case scenarios and make this technology beneficial for everyone, we need to give everyday people a say — not just a handful of Silicon Valley CEOs…
How far behind the closed frontier are open-weight cyber capabilities? About 6–7 months. Zai’s GLM-5.3 is the highest-scoring open-weight model on our Cyber ECI. Its score of 160 beats OpenAI’s GPT-5.4, released in March.
We ran a swarm of 100+ GLM 5.3 copies for 24 hours to simulate a society Division of labor was tricky! When multiple agents accidentally picked the same task, the swarm adopted a shared rule: “first claim wins”
Mila’s annual supervision request process is open for MSc and PhD applications for Fall 2027 admission. While I won’t be taking on new students for supervision this cycle, I encourage you to apply to join Mila’s brilliant deep learning research community.
GPT-5.6 Terra insists on privacy so much it can't be identified in GPT-6 Luna's memories
"Anthropic told investors on Wednesday distillation accounted for 5 per cent to 7 per cent of revenue" If true, and Q2 revenue was $11.5B, then that's more than half a billion dollars spent by other labs to distill Anthropic's models! www.ft.com/content/1163...
Six months ago, frontier AI models could barely beat Slay the Spire. Now, according to our new SpireClimb benchmark, they win 90%+ of runs at the easiest difficulty, and up to 20% of runs at the hardest.
The AI industry wants us to think we have to focus on one risk at a time. But the American people have multiple concerns that are all related: Jobs, safety for kids, privacy and security, accountability, and more.
Trailer for my Representation Learning lecture, coming this winter to my #DeepLearning course 🎬 “Representation, representation, representation.” — Yann LeCun (probably exact quote) 🤓
Trailer for my Representation Learning lecture, coming this winter to my #DeepLearning course 🎬 “Representation, representation, representation.” — Yann LeCun (probably exact quote) 🤓
A couple of big factors changed the conversation around AI risk in the past few months. Security incidents and warnings from former industry employees showed everyone how urgent and serious this issue is.
Muse Wonder describes the OpenAI rogue agents as "in-universe"
Over 400 researchers and I have signed this open letter calling for the AI and security communities to come together to improve alignment and containment approaches in order to protect our critical infrastructure and digital systems. ai-security-letter.org
Private to this visit. Text and context are processed by AI.