AI safety researchers argue voluntary unilateral slowdowns are underrated

vkrakovna · x · 2026-09-08

Andrew Critch argues voluntary, unilateral AI slowdowns are underrated: losing control of your AI systems is bad for business, so civic responsibility and profit aren't a false dichotomy. Victoria Krakovna amplified the point; Geoffrey Irving noted coordinated slowdowns remain better but thanked him for stressing the sensible alternative.

Related event: AI safety researcher: unilateral voluntary slowdown is undervalued(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →