AA_CT_3 Refusal Termination and Non Continuity in Artificial Agent Governance

This paper establishes a structural result concerning refusal and termination in artificial systems: any refusal or termination event necessarily breaks standing continuity and therefore cannot be treated as an internal state change of a single, persisting artificial agent. It proves that refusal, shutdown, reset, or enforced cessation constitute standing-terminating events rather than admissible transitions within an agent’s identity. The result is eliminative and non-constructive. The paper does not propose governance mechanisms, refusal policies, safety protocols, or deployment practices. Instead, it shows that treating refusal or termination as a governed behavior of an enduring agent is inadmissible, because standing cannot be preserved across such events. Any system resumed after refusal or termination is shown to be a numerically distinct instantiation, regardless of architectural similarity, memory retention, or functional continuity. The analysis blocks a class of claims in which artificial agents are said to “choose” refusal, to remain accountable across shutdown, or to preserve identity through interruption, retraining, or restart. It demonstrates that these claims rely on illicit continuity assumptions and collapse into surrogate notions of agency once standing preservation is examined.