AI 2040 Plan Proposes Delaying Superintelligence Until Safe

geoffreyirving · x · 2026-07-29

Safety researchers published AI 2040: Plan A, a positive vision to avoid AI-driven existential catastrophe. The plan recommends an international deal to delay superintelligence (SI) development until 2040 and mandate total research transparency. It proposes a regime of mutually assured compute destruction, allowing multiple companies globally to scale slowly and safely together instead of racing in secrecy.

Related event: Researchers Propose Delaying Superintelligence to 2040(3 posts)→

Original post →

More from AGI Musings

AGI Musings channel →