Ex-OpenAI researcher: voicing AI safety concerns was unpopular inside the lab

nabla_theta · x · 2026-09-11

nablatheta responds to @andrewho03, saying their OpenAI experiences differ sharply: until very recently, being outspoken about AI safety and existential risk meant constant disagreement and unpopularity, with incentives pushing against thinking hard about it.

He adds that many people downplayed their level of concern for fear of social perception — a window into safety discourse culture inside a frontier lab.

Related event: Insiders Say Questioning AI Safety Is Unwelcome Inside Frontier Labs(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →