AI 2040: Is it Actually a Deal?

On Monday, the AI Futures Project released AI 2040: Plan A, a proposal for governing powerful AI systems by twenty forty. Per LessWrong, the core idea is ambitious: give governments total visibility into all AI compute use and establish a mutual-assured-destruction capability between the US and China, where each nation could disable the other's AI infrastructure if safety standards are breached. But the plan has a fatal flaw—it outlines how to build doomsday buttons without specifying what happens when the US and China disagree on whether some AI training method is safe. The proposal says they bargain and threaten, then one side caves. The critic argues that real international agreements don't just provide transparency for power plays; they establish procedures that constrain disputes away from mutual annihilation. Without such guardrails, why would any nation accept a deal where disagreement means the destruction of major segments of its economy?

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton