Allie Miller's checklist for AI quitters: name real risks, not vague panic

alliekmiller · x · 2026-09-17

Reacting to AI employees loudly quitting over safety concerns, Allie Miller laid out what they should actually say: how long they worked in AI and worried about what specifically (misalignment, bioweapons, coordination), their timelines and confidence levels, concrete failure modes ordinary people can picture, and real policy asks — data center environmental standards, third-party lab evaluators, whistleblower protections, required public risk disclosures before releases, chip export controls — instead of vague "AI will kill us all" panic. She also suggests having non-engineers, non-tech people, other generations and at least one AI optimist review the statement, contrasting with how Hinton handled leaving Google.

Original post →

More from AGI Musings

AGI Musings channel →