Reconstructing what an AI system did is a forensic capability. Proving what was authorised before it acted is an architectural one. The governance documentation market — law firms, consulting practices, compliance frameworks — sells the first. Enforcement proceedings under AI Act Article 99 will demand the second. This article establishes the structural distinction between evidence-based governance and proof-based governance, and why only one survives adversarial scrutiny.

Governance only becomes enforceable when three conditions are simultaneously met: T‑0 fixation, structural independence (S ∩ C = ∅), and legal opposability. Without these properties, no post-execution record can establish a governance state that was never sealed. This article articulates the architectural conditions under which governance ceases to be a management assertion and becomes a verifiable system property.