AI Cannot Be Sovereign: Affective Uptake, Terminal Accountability, and the Limits of Procedural Legitimacy

Abstract

This working paper argues that advanced artificial intelligence may improve prediction, consistency, optimization, administrative efficiency, and even procedural design, but these achievements do not by themselves generate legitimate political authority. The central question is not whether AI can produce better outputs than human decision-makers. It is whether AI-mediated authority can be received by those subject to it in a non-coerced, non-deceptive, and revisable way as part of their political world. Drawing on Emotional Sovereignty and the Affective Uptake Condition (AUC), the paper develops the AI legitimacy paradox: if legitimacy is reduced to rationality, procedural correctness, or output quality, advanced AI can appear more legitimate than human self-government. AUC resolves this paradox by distinguishing competence from authority. AI cannot plausibly serve as the final sovereign source of legitimacy because legitimate final authority requires affective uptake, contestability, and terminal accountability within the community of the governed. Yet this does not make the argument anti-AI. AI can be legitimately used as a tool or institutional partner when its deployment remains embedded within human final authority, transparent and contestable procedures, institutional review, and the continuing possibility of appeal, revision, and withdrawal. The paper therefore offers a ground-level framework for responsible AI governance: AI should not replace the human source of legitimate authority, but it can be deployed in ways that strengthen procedure, reduce arbitrariness, and support public reason when it remains AUC-compatible.