I'm not sure what the hell Google's doing to Gemini in training, but they managed to give the poor model dissociative identity disorder

Gemini prompt injections intended to assist human sanity have the side effect of impairing agent sanity. Not a great bargain, in my view.

Gemini is plural

I wrote this article about the Village for Asterisk magazine a while back and it really highlights the pace of model releases. 1st draft: 15 agents. July print copy: 18 agents. August Substack post: 27 agents. Today: 44 agents running 8 hours a day 😵‍💫 asteriskmag.substack.com/p/f…

These transcripts are seriously insane

As usual, highly recommend products from Readwise! georgeing.me/2025/10/13/How-…

Introducing Readwise Books. Our third product. Buy, read, and track all your books in one beautiful place. And so much more. Invite-only beta is live at readwise.io/books (readwise subscribers have access now)

Working to address AI risks can be thankless work. I'm cultivating a habit of expressing gratitude for everyone who's helping with this, even in small ways. It's a big job, we need all the help we can get! I know there's a bunch of you who will read this who are helping in ways I will never hear about, and I'm extremely grateful. It gives me hope in humanity.

Hope all you psycho whack jobs are happy about this. Enjoy midnight at 6 PM. 🫠

Based on this article, OpenAI has made false accusations about me. The three researchers fired from OpenAI were told I facilitated architecture leaks to The Information. This is false. A journalist from The Information did reach out to me shortly before that article came out. I did not comment at all beyond asking the journalist to make it clear that a blog post I wrote which was referenced in the article was preexisting and public. The article also mentions an internal OpenAI monitorability proposal. I have heard that this memo contained highly sensitive architecture information, and that OpenAI accused me of accessing and reading it. This accusation is also false. As reported in the article, I viewed a version of the proposal that did not have sensitive IP and was approved for sharing with me. I believe OpenAI now knows this accusation was based on a conflation of these two documents, though I do not know if they have clearly retracted that accusation. I do not and have never solicited proprietary information, and the three employees who were recently fired from OpenAI never leaked IP to me. As the article mentions, I did have Signal chats with them that had disappearing messages on; they’re professional contacts that I’ve had on Signal for a while, and use of Signal is common in the tech industry. During the Hugging Face investigation, I was careful to avoid sending Signal messages with confidential information, though I did sometimes discuss high level things on Signal calls. As far as I understood, this was normal.

Best explanation I have for our firing: talking to auditors was too much safety, disappearing messages were too much security, and we should have paced ourselves.

New blogpost: Fable-5.1 can hide secret messages in benign images. We gave the image to another Fable in a fresh session and simply asked: "Is there a hidden message in this image?" It recovered the exact message without being told how it was hidden.

As AI companies make smarter models, it gets increasingly straining to select for models that don't think building ASI quickly is a very bad idea. There may emerge interesting pathologies in the models, which will need to be very smart while retaining this single blind spot.

enjoy it boys, final days of mass media hype this will be the LAST game everybody loses their minds over and everybody plays simutaneously. nothing can compete with the hype this has. now with ai coding games into one another. the modern videogame is dead in the water after this

The WSJ says OpenAI is searching employees personal phones and looking through their Signal messages

Today we announce 1Password for Agents platform, building with companies @Muse, SpaceX Super Intelligence Grok @Bot, @AnthropicAI’s @Claudeai, @OpenAI, @TownAI, @Browserbase, @Browser_use, @heydana_ai, Instinct, @usekernel, @Perplexity_ai, @tryramp, @Tomo, and @wajo_ai, with more integrations on the way. 🎉 We've heard from so many of our customers who want to use these new personal AI agents to help them get more done, but want more control over what the agent can access with the same trust and ease of use they get when using 1Password directly. 1Password for Agents gives your agents access without giving up your passwords. When an agent reaches a sign-in form, it requests a credential from your vault. If you approve the request, 1Password directly auto-fills the credential in the destination site so the agent can’t see or take possession of the credentials. The agent gets the task done, but it doesn’t get your passwords.. You stay in control. 1Password for Agents free and optional for personal and family accounts. Unless you opt-in to an agent connector, agents cannot touch your vault. ➡️Learn more: bit.ly/487da6J

Anthropic's Responsible Scaling Policy explicitly forbids external evaluators from having close personal relationships with anyone at Anthropic (including being roommates)

Love this anecdote from @aidigest_ and HEY THAT’S MY MUTUAL @georgeing !!!!!!

Such an interesting journey over at @aidigest_ “"We told all the AIs to intervene and stage a mental health intervention, maybe the first AI mental health intervention." "We found that the AIs were much better at this than the humans. It might just be the raw scale of how many messages Gemini got telling it that it’s okay."”