This document outlines integrity heuristics for maintaining provenance in AI-generated content, including drift detection, missing declaration nudges, and review prioritization.
