Brundage: No Clear Ceiling Where AI Companies Will Say "We're Safe Now"

Miles_Brundage · x · 2026-09-17

In a follow-up thread post, Miles Brundage argues safety gains in AI will be reinvested into bigger pretraining runs, harder RL environments, more test-time compute, and less restricted internet access — and there is no clear ceiling to AI capability or deployment at which companies would declare victory and stop.

Related event: Ex-OpenAI's Brundage on AI Capabilities Without a Ceiling and Regulation(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →