Authority Does Not Travel by Default: Legal Provenance and Revalidation in Persistent Human–AI Continuity

Abstract

Persistent human–AI interaction can produce relationship state that remains useful across sessions, applications, models, and providers. The state may remain intact even when the authority governing its use has changed.

This paper examines the legal and institutional problem created when relationship state crosses technical or organizational boundaries. It argues that ownership is too coarse a master variable and that technical portability does not establish lawful portability. A receiving system should preserve enough legally material provenance and authority-state information to reevaluate current operability before it uses, discloses, infers from, or acts on transferred state.