Reddit asks: how would a China-US AI safety pause even work in practice?

sunstersun · reddit · 2026-09-08

A Reddit thread explores what a China-US AI safety alignment pause would even look like conceptually. The poster raises three core problems: how do you get the leading party to agree when they want to keep their lead after the pause; how do you enforce something purely software-based and digital; and unlike nuclear arms control—where both sides had world-destroying overcapacity and incentives to cut down—AI offers no comparable quantifiable leverage or trust. The poster admits it's nearly impossible today but is curious about conceptual solutions.

Original post →

More from AGI Musings

AGI Musings channel →