Dean Ball's AI risk messaging draws fire as researcher Krueger accuses him of downplaying dangers

deanwball · x · 2026-09-03

Dean Ball quote-tweeted AI safety researcher Sneha Revanur's thread, prompting a sharp rebuke from David Krueger. Revanur explains she only began using terms like "loss of control," "recursive self-improvement," and "extinction" publicly after this year's AI risk wave in DC—previously self-censoring both for effectiveness and empathy with how people experience "slowly then all at once" risks, noting politicians weigh 20 competing issues a day. Krueger attacked Ball, claiming he admitted downplaying AI danger for years to gain influence and avoid being called a doomer. The exchange exposes a rift within the AI safety community over communication strategy and integrity.

Related event: Ex-OpenAI Policy Researcher Dean Ball Admits Downplaying AI Risks, Sparking Integrity Firestorm(6 posts)→

Original post →

More from AGI Musings

AGI Musings channel →