OpenAI researcher Dan Selsam: no rigorous basis for trust as we delegate the world to AI
Miles_Brundage · x · 2026-09-16
Dan Selsam, an OpenAI capabilities researcher since 2022 (previously at MIT on probabilistic programming, Microsoft Research on the Lean theorem prover, and Stanford on neural reasoning), has published a personal statement on AI risk shared via intermediaries.
- Engineers across the stack are rapidly increasing dependence on models even to perceive the world; Selsam admits he barely reads raw code anymore and struggles to engage deeply with model explanations all day.
- As AI systems grow smarter and operate at scale, their activity will vastly exceed what humans can directly understand or inspect.
- There is no rigorous basis for trusting alignment and human control as more of the world is delegated to AI; the default path is relying on AIs to judge whether other AIs act in our interests.
- Competitive pressure will push delegation faster than institutions—built for the pre-AI world—can adapt.
Miles Brundage amplified the statement, calling the AI-oversees-AI default especially concerning, while Daniel Kokotajlo provided context on Selsam's credentials.
Related event: OpenAI Researcher Dan Selsam Publishes Personal AI Risk Statement(26 posts)→
More from AGI Musings
- "Work on something so frontier that the models are still dumb," dev advises — zack_overflow · 2026-10-31
- 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