Salib clarifies: rogue AI risk and social-science framings are distinct claims

petersalib · x · 2026-09-08

Wrapping the debate, petersalib clarifies: one can disagree that rogue AI, rogue labs, or rogue governments using AI are big risks modulated by capability progress — but that is a distinct claim from 'we can use social science to understand AI,' which he agrees with. The thread narrows to a dispute over claim boundaries rather than substance.

Related event: AI safety debate: jailbreaking as model property vs human operations failure(8 posts)→

Original post →

More from Safety

Safety channel →