Geoffrey Irving Asks: What Level of Misalignment Accident Would Change the Calculus for Unilateral Stops?

geoffreyirving · x · 2026-08-05

Geoffrey Irving tweets that his idea that it might be rational for a lab to unilaterally stop has received mostly pushback. He suggests thinking through what level of dramatic misalignment accident would change that calculus and invites replies.

Related event: Geoffrey Irving Discusses Alignment Incidents That Could Halt AI Labs(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →