Cognitive Jurisdiction: Formal Limits on Artificial Intelligence Participation in Human Decision Domains

Abstract Recent advances in artificial intelligence have intensified concerns not only about accuracy, alignment, or safety, but about the scope of legitimate participation of artificial systems in human decision-making. While the theory of Governed Cognition establishes the right to reason as a conditional and enforceable constraint on artificial intelligence, it leaves open a critical structural question: where is an artificial system permitted to participate at all? This paper introduces Cognitive Jurisdiction, a formal framework that defines territorially bounded domains of legitimate artificial intelligence participation. Cognitive jurisdiction separates domains of permitted reasoning, shared assistance, and absolute prohibition, independent of system capability, accuracy, user consent, or safeguards. The framework formalizes jurisdictional boundaries, jurisdictional overreach, and mandatory abstention as first-class governance constructs. We argue that many contemporary AI risks—including authority simulation, responsibility displacement, and existential fear narratives—arise not from insufficient alignment or control, but from ungoverned jurisdictional participation. By denying artificial systems the right to enter human-exclusive decision domains, Cognitive Jurisdiction renders entire classes of feared AI behaviors structurally impossible rather than merely undesirable. This work extends Governed Cognition by providing a jurisdictional layer that governs where cognition may occur, complementing prior constraints on whether and how reasoning may proceed. The result is a complete, auditable model of authority-aware artificial intelligence suitable for high-consequence and regulated environments.