If a regulator asked your board tomorrow what your AI did and why, could you produce a defensible answer? Most boards can't. AI has moved from an IT project to a system that reasons, decides, and acts on behalf of the enterprise — but governance hasn't followed it into the boardroom. This article makes the case that boards must stop governing AI like a technology asset and start governing it like an autonomous decision-maker with its own blast radius, accountability trail, and structurally enforced boundaries.