# Operator Sovereignty **Domain:** Human Agency / AI Governance / Accountability **Doc Type:** Canonical Concept Node **Maturity:** Evolving **Related:** [[Human Deliberation]], [[Human Agency]], [[Moral Accountability Layer]], [[Advisory Constitution]] --- ## Definition **Operator Sovereignty** is the requirement that a human or plural decision layer retain genuine authority to question, refuse and redirect machine recommendations while remaining accountable for the resulting action. ## More Than a Human in the Loop An operator is not sovereign when refusal is unavailable, punished, too late or impossible to explain. Nor does operator sovereignty permit humans to use machine advice as a liability shield. The necessary conditions include access to reasons and uncertainty, time, lawful alternatives, independent evidence, control over execution and review of both acceptance and refusal. Where no individual can safely bear the decision, authority should move to a plural institution rather than back into the machine. Finch's refusal in [[Person of Interest S3E20 — Death Benefit]] is the paradigmatic case. It preserves moral jurisdiction even under catastrophic uncertainty. [[Person of Interest S5E11 — Synecdoche]] shows that operator authority can be distributed across teams rather than concentrated in one creator. ## Key Insight **A human remains accountable only if the human retains a real decision.** ## See Also [[Selective Disclosure]], [[Functional Privilege]], [[Constitutional Safeguards]], [[Prediction Is Not Jurisdiction]], [[Human-Machine Symbiosis]] ## Sources / Provenance - Derived from [[wiki/Person of Interest|Person of Interest]], [[Human Deliberation]] and [[Moral Accountability Layer]].