Ethan Mollick: has a frontier model actually caused a major security incident in normal deployment?

emollick · x · 2026-09-23

Ethan Mollick asks a pointed question: organizations were deeply worried in 2023 about frontier AI models causing major security incidents in internal commercial use, yet he can't recall any real incident occurring in ordinary deployments with normal safeguards. He invites counterexamples, implicitly challenging whether frontier-model risk narratives match the empirical record.

Original post →

More from AGI Musings

AGI Musings channel →