AI safety researcher: coordination window closes around 2028 once RSI becomes possible

JeffLadish · x · 2026-09-07

AI safety researcher Jeff Ladish warns that humanity has a critical window for coordination before recursive self-improvement (RSI) becomes possible — and that window closes shortly after, around 2028 if OpenAI's timeline holds.

He adds that while it's hard to grapple with a life-and-death situation and stay clear-headed, ignoring it doesn't make it go away. Labs aren't stopping despite many employees wanting to: they're building automated researchers.

Related event: Researcher Warns AI Coordination Window May Close by 2028(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →