Taxonomy of Inference-Time AI Governance and Monitoring Mechanisms

This framework proposes twenty inference-time mechanisms for monitoring, verification, and enforcement as model capabilities shift from training to deployment. It evaluates readiness against a four-vendor evidence base and tests mechanisms against various adversary models.