RT @MariusHobbhahn: I'll be a mentor for the Astra Fellowship this round. Come join me to work on better black box monitors for scheming!….

RT @woj_zaremba: It’s rare for competitors to collaborate. Yet that’s exactly what OpenAI and @AnthropicAI just did—by testing each other’s….

Join millions who have switched to Grok.

RT @ThankYourNiceAI: No single person or institution should define ideal AI behavior for everyone. Today, we’re sharing early results fro….

RT @hauntsaninja: i sampled some of OpenAI's older models; i think this helps you feel AI progress more viscerally

RT @apolloaievals: We've evaluated GPT-5 before release. GPT-5 is less deceptive than o3 on our evals. GPT-5 mentions that it is being e….

RT @tszzl: we've been testing some new methods for improving writing quality. you may have seen @sama's demo in late march; GPT-5-thinking….

RT @woj_zaremba: Red teamers assemble! ⚔️💰. We're putting $500K on the line to stress‑test just released open‑source model. Find novel risk….

RT @OpenAI: We’re launching a $500K Red Teaming Challenge to strengthen open source safety. Researchers, developers, and enthusiasts world….

RT @boazbaraktcs: I didn't want to post on Grok safety since I work at a competitor, but it's not about competition. I appreciate the scie….

RT @balesni: A simple AGI safety technique: AI’s thoughts are in plain English, just read them. We know it works, with OK (not perfect) tra….

Really recommend this post on scheming. I hadn’t really understood the potential complexities of scheming until reading (an earlier preview) — especially the last points about deceptive alignment and situational awareness.

RT @MilesKWang: We found it surprising that training GPT-4o to write insecure code triggers broad misalignment, so we studied it more. We f….

RT @aidan_mclau: i’m forming a model behavior research team !. i'm realizing llms are not alien artifacts, but rather craftsman works, like….

RT @joannejang: some thoughts on human-ai relationships and how we're approaching them at openai. it's a long blog post --. tl;dr we build….

I don’t agree with all of the takes but found this to be a really interesting and thought provoking conversation (even as someone who has been following METR’s work relatively closely).

RT @MariusHobbhahn: LLMs are getting rapidly more evals aware!. Afaik, nobody has a good plan for what to do when the models constantly say….

This captures well a lot of what worries me most about AI. The logical result of unfettered capitalism + smart AI is to take capitalism to its extreme — “labor” becomes capital too. And probably a really well functioning political system could help ease the transition, but….

Well done.

RT @dlwh: Come read about all the mistakes I made along the way to beating Llama 3.1 8B on 14/19 benchmarks. We trained from scratch, made….

RT @polynoamial: It's deeply concerning that one of the best AI researchers I've worked with, @kaicathyc, was denied a U.S. green card toda….