The Compound Threat Matrix: Structural Vulnerabilities in AI Systems Requiring Unified Governance Infrastructure

This compendium documents over 150 verified incidents, regulatory actions, peer-reviewed findings, and quantified impacts across six functional domains of AI deployment between 2024 and February 2026. Its original contribution is the identification of compound interaction pathways through which vulnerabilities in one domain — evaluation integrity, model supply chains, operational deployment, information systems, infrastructure, and governance — propagate into and amplify failures in others. Five primary cascade pathways are traced with confirmed evidence at every step. A full domain-by-domain interaction matrix demonstrates that no domain is isolable: every domain both transmits failures to and receives failures from at least four others. The analysis establishes that governance architectures addressing AI risk category-by-category are structurally incapable of containing risks that propagate across category boundaries, and derives five minimum structural requirements for any adequate governance response: hardware-rooted trust, continuous behavioral monitoring, supply chain provenance binding, binary compliance architecture, and composition-based standards.