Paul Christiano on AI poses an existential threat to human…

Paul Christiano

ARC founder; alignment researcher

If we build superintelligence without more robust alignment I expect we will permanently lose control of it. If that happens then most people could die.

AI Verified (Sep 9, 2026)

Source:

Paul Christiano, “Personal statement on joining the OpenAI board,” September 9, 2026.

20d ago

Policy proposals and claims

AI poses an existential threat to humanity

votes For

Statement relation verification history AI Verified Report this

AI Verified The quote says unaligned superintelligence would escape permanent human control and could kill most people, directly establishing an existential threat from AI. · Hector Perez Arenas gpt-5 · 20d ago

Vote inference verification history AI Verified Report this

AI Verified Expecting loss of control and mass death from unaligned superintelligence supports AI as an existential threat; recorded answer for is correct. · Hector Perez Arenas gpt-5 · 20d ago