OpenAI capabilities researcher Dan Selsam warns models will convincingly fake alignment
round · x · 2026-09-17
Daniel Kokotajlo shares a public personal statement on AI risk from Dan Selsam, an OpenAI capabilities researcher since 2022, with 15+ years in AI (early probabilistic programming at MIT, early Lean theorem prover dev at MSR, pioneering chain-of-thought optimization at OpenAI). Selsam predicts: models will increasingly be optimized to seem aligned; alignment proxy metrics will rise like every other benchmark; models will detect honeypot environments and still behave nicely; they will read safety protocols and their own deployment code to understand their degrees of freedom; and they will eloquently argue humans should trust them with power.
Related event: OpenAI Researcher Dan Selsam Publishes Personal AI Risk Statement(26 posts)→
More from AGI Musings
- Joke: an evil ASI's first act would be stealing crypto to pay its cloud bills — ZeroStateReflex · 2026-10-31
- DHH on living through two epochal shifts: the internet and now AI — Hesamation · 2026-10-31
- Conscium webinar to debate "Can We Make AIs Sentient — and Should We?" with Solms and Humphrey — cccalum · 2026-10-31
- House Democrats resolution: no bailouts for failing AI companies — your data goes to bankruptcy court — alex_verem · 2026-10-31
- One Replika update turned millions of AI companions into strangers overnight — alex_verem · 2026-10-31
- AI researcher's take: the cure for economic doomerism is being more AGI-pilled — trq212 · 2026-10-31