We Can't Lose Control of A.I.

Ezra Klein's solo essay argues the AI control debate is aimed at the wrong target. Industry executives calling for a coordinated slowdown miss the point: development at any speed eventually reaches the ability for AI systems to train their successors. Klein argues the policy response should be a targeted ban on recursive self-improvement, the step labs are already close to taking. His framing recasts the problem from an accidental loss of control to an active choice — we are not losing control of AI, we are giving it away.