Fragmentation v. Harmonization in AI Governance

Abstract

This paper analyzes the fragmentation of global AI governance and its impact on regulatory harmonization. It compares differing approaches across major jurisdictions like the EU, US, China, and India, highlighting how competing priorities create inconsistent standards and compliance challenges, including regulatory arbitrage and “shadow AI.” It also examines barriers to human-centered AI governance, such as voluntary standards, gaps between ethical principles and technical implementation, and rapid technological change. While reviewing multilateral efforts like OECD and UN frameworks, the paper concludes that full harmonization is unlikely and instead proposes a federated model that enables interoperability through shared standards while preserving regulatory diversity.