AI Integrity: A New Paradigm for Verifiable AI Governance

arXiv:2604.11065v1 Announce Type: new Abstract: AI systems increasingly shape high-stakes decisions in healthcare, law, defense, and education, yet existing governance paradigms -- AI Ethics, AI Safety, and AI Alignment -- share a common limitation: they evaluate outcomes rather than verifying the reasoning process itself. This paper introduces AI Integrity, a concept defined as a state in which the Authority Stack of an AI system -- its layered hierarchy of values, epistemological standards, so