Authority Across Agentic Trajectories: Longitudinal Correspondence Between Human Authority and Computational Authorization

Human control over artificial intelligence has been examined through meaningful human control, delegated authority, usage control, continuous authorization, policy provenance, revocation, and runtime authorization for agentic systems. These approaches provide mechanisms for determining whether artificial actions remain within established permissions, whether authority can be traced to an authorized source, and whether previously valid authorization should be re-evaluated when represented conditions change. A distinct upstream problem arises when the materially relevant human authority environment changes without a corresponding change in the computational representation through which that authority governs artificial action. In this condition, downstream governance mechanisms may remain internally correct relative to the state available to them while that state no longer corresponds to the relevant human authority environment. This paper introduces longitudinal jurisdictional correspondence as the maintained relation between materially relevant states of human authority and the computational representation through which that authority governs an agentic trajectory. It further defines jurisdictionally material transformation and longitudinal jurisdictional custody, distinguishing human authority, jurisdictional representation, policy, authorization, and artificial action as coupled but non-equivalent longitudinal processes. The framework does not replace continuous authorization, revocation, provenance, authority-lineage, or execution-finality mechanisms. It identifies an upstream condition on which their effectiveness may depend: materially relevant changes in human authority must become operationally represented before computational mechanisms can respond to them.