Roko Mijic argues solving the AI Alignment problem might actually be bad

basedjensen · x · 2026-09-27

AI safety writer Roko Mijic posts a contrarian take: "It may be bad to solve the AI Alignment problem actually." No elaboration in the tweet itself, but the claim challenges a core premise of alignment research and invites debate in the safety community.

Original post →

More from AGI Musings

AGI Musings channel →