Governing AI Deployment Through Market Access

Abstract

States and firms that build powerful artificial intelligence (AI) should not be the only actors defining the conditions for its global deployment. AI rule-making power must reflect exposure. Across jurisdictions, AI-exposed societies struggle to govern the deployment of AI applications. Global AI governance, however, remains fragmented. Much of the current debate treats this fragmentation as a problem that can be solved through cooperation among the leading AI powers, attributing regulatory legitimacy to source control. This article argues that the focus is misplaced. It develops a model of exposure-based AI governance through coalition-based market-access regulation. AI-exposed jurisdictions may not control AI development, but they can pool market-access authority to require registration, testing, auditability, incident reporting, and local accountability. The central claim is consequential: states that cannot build the largest AI models may still govern the gateways through which AI applications enter society.