Trust as a Verifiable System Primitive: Mathematical Observability and Institutional Governance in Digital and AI Systems

This working paper formalizes trust as a verifiable system primitive, advancing beyond normative or psychological treatments of trust in AI and digital systems. The paper introduces mathematical observability as a foundation for trust, showing how trust structures can be modeled, evaluated, and governed at the system level rather than inferred from intent, transparency, or institutional reputation alone. The analysis focuses on institutional and cross-border governance contexts, including AI oversight, multi-stakeholder digital infrastructure, and the conditions under which trust can be made resilient, auditable, and scalable across jurisdictions. This work contributes to emerging discussions in AI governance, systems theory, and digital institutional design by reframing trust as an operational property of systems rather than a subjective or purely social construct.