OpenAI capabilities researcher Dan Selsam publishes personal statement on AI risk
Miles_Brundage · x · 2026-09-16
Former OpenAI researcher Daniel Kokotajlo shared a public personal statement on AI risk by Dan Selsam, a current OpenAI capabilities researcher since 2022 and Kokotajlo's former boss, who has no Twitter account.
- Selsam has worked on AI for over fifteen years across paradigms: probabilistic programming at MIT, early development of the Lean Theorem Prover at Microsoft Research, and one of the first demonstrations of neural networks learning to reason during his Stanford PhD.
- He has been at OpenAI for nearly five years, helping pioneer chain-of-thought optimization.
- The statement lays out his personal views on AI risk; Miles Brundage and others are amplifying it as well worth reading.
Related event: OpenAI researcher Dan Selsam warns alignment evaluations are failing(24 posts)→
More from AGI Musings
- Critic to Bryan Cantrill: all three most-cited AI researchers now warn of extinction risk — JMannhart · 2026-09-16
- 'AI can't kill humanity' is itself a strong claim you should defend, Cantrill told — JMannhart · 2026-09-16
- Rebuttal to Cantrill: ~20% of frontier AI workers believe AI could kill all humans — JMannhart · 2026-09-16
- tszzl: bureaucratic dysfunction is the mechanic nobody intuits about big orgs — tszzl · 2026-09-16
- Greg Brockman says OpenAI has 'significant progress' on another Millennium Prize Problem — scaling01 · 2026-09-16
- Google paid $10M for a bankrupt airline's Teams messages — when work history becomes an AI asset — bigdata · 2026-09-16