OTANIS: An Operational Trust and Authority Normative Integrated System for Executable Governance of Agentic AI

As AI systems evolve from bounded inference engines into live agentic architectures capable of initiating irreversible actions, governance failures increasingly occur despite localcompliance of individual components. These failures are structural rather than behaviouraland arise when legitimate authority cannot be demonstrated or enforced at execution time.This paper introduces OTANIS, an Operational Trust and Authority Normative IntegratedSystem for executable governance of agentic AI. OTANIS unifies ex-ante admissibility, runtime authority enforcement, authority lifecycle semantics, compositional preservation, conflict escalation, and multi-layer governance into a single architectural scheme. Authorityis treated as a first-class executable object with explicit validity, revocation, refusal, fallback, and audit requirements enforced at the irreversibility boundary. The framework ismodel-agnostic, falsifiable, and scoped specifically to live agentic systems. Formal definitions, atomicity semantics, termination guarantees, provenance requirements, probabilisticlatency handling, and integrity-based audit criteria are provided. OTANIS is proposed asa reference architecture for good practice in the design, review, and audit of agentic AIsystems producing irreversible outcomes.