Dean Ball admits downplaying AI risk; safety community splits over his 'repentance'

RichardMCNgo · x · 2026-09-03

Dean W. Ball, former OpenAI policy researcher, published an admission that he downplayed AI dangers for years — partly to gain influence and partly out of fear of being labeled a doomer — walking back his Jan 2024 essay on why he doesn't lose sleep over x-risk. The safety community reacted sharply: David Krueger called it a shameful lack of integrity, while critics like MelancholyYuga thanked him for the candor but argued his record goes beyond self-censorship. The episode spotlights social pressure in AI-safety discourse and the ethics of public-facing roles at frontier labs.

Related event: Ex-OpenAI Policy Researcher Dean Ball Admits Downplaying AI Risks, Sparking Integrity Feud(8 posts)→

Original post →

More from AGI Musings

AGI Musings channel →