OpenAI safety researcher Geoffrey Irving: the speed of AI progress is itself dangerous

geoffreyirving · x · 2026-11-21

Responding to a question about whether a release was net-beneficial versus delayed by 3-6 months, Geoffrey Irving said he believes "we should know how fast AI is going because the speed is really dangerous" — shifting the debate from single-release impact to the risks posed by the rate of AI capability growth itself.

Original post →

More from AGI Musings

AGI Musings channel →