OpenAI researcher Dan Selsam publishes personal statement on AI risk, peers push back
joshua_saxe · x · 2026-09-15
Dan Selsam, an OpenAI capabilities researcher since 2022, released a public personal statement on AI risk, shared on his behalf by former boss Daniel Kokotajlo (Selsam has no Twitter account).
His background: 15+ years in AI across paradigms — early probabilistic programming work at MIT, early Lean Theorem Prover developer at Microsoft Research, one of the first demonstrations of neural networks learning to reason in his Stanford PhD, and co-pioneering chain-of-thought optimization at OpenAI.
The statement argues that if a powerful model ever escapes human constraints, it likely won't stay within intended bounds, and its unchained behavior could plausibly cause runaway industrialization making the planet inhospitable. Commenter herbiebradley agrees with much of it but says it makes the classic mistake of assuming "intelligence is all you need," calling it a category error about the world; joshuasaxe reposted the critique.
More from AGI Musings
- Should AI models pass NIST-style safety checks before shipping? Industry voices say it's simple — aronchick · 2026-09-15
- Physicist Sabine Hossenfelder on arXiv and AI papers: flag them, don't ban them — skdh · 2026-09-15
- Common AI forecasts ignore algorithmic breakthroughs, argues one skeptical poster — shauseth · 2026-09-15
- Yacine: AI can't get much smarter — only faster from here — yacineMTB · 2026-09-15
- "Name One Thing AI Actually Made Better": Dev's Viral Skeptical Rant Beyond Coding — sebpaquet · 2026-09-15
- Analysis: Anthropic pacing alone buys one month; coordinated US slowdown cuts risk to 21% — soleio · 2026-09-15