CONVERGENCE OR FRAGMENTATION? A COMPARATIVE ANALYSIS OF AI GOVERNANCE FRAMEWORKS IN THE EU, UNITED STATES, AND CHINA
DOI: https://doi.org/10.56815/IRJAHS.v(2026)i2.1-10
Keywords:
AI governance; EU AI Act; algorithmic regulation; comparative policy analysis; regulatory fragmentation; AI executive orders; China AI regulation; global AI governance; regulatory harmonization; risk-based regulation.Abstract
As artificial intelligence systems increasingly transcend national borders, the regulatory frameworks governing them have developed along strikingly divergent paths. This paper undertakes a comparative analysis of three of the world's most influential AI governance regimes: the European Union's AI Act, the United States' executive order-driven approach, and China's evolving suite of AI regulations. Using a structured comparative framework, the study examines each jurisdiction's regulatory philosophy, risk classification methodology, enforcement mechanisms, and treatment of general-purpose versus narrow AI systems. The EU's AI Act represents a comprehensive, risk-tiered, rights-based model rooted in ex-ante compliance; the US approach reflects a more fragmented, sectorspecific, and executive-driven strategy shaped by shifting political priorities and limited congressional action; China's framework combines state-driven algorithmic oversight with strategic industrial policy objectives, emphasising content control and social stability alongside innovation goals. The paper argues that while surface-level convergence is emerging around certain technical concepts — such as risk categorisation, transparency obligations, and safety testing — the underlying regulatory logics, enforcement capacities, and normative priorities remain substantially fragmented. This divergence carries significant implications for multinational AI developers navigating compliance across jurisdictions, for the prospects of international regulatory harmonisation, and for the broader question of whether a global AI governance regime is achievable or desirable. The paper concludes by evaluating pathways toward interoperability, including mutual recognition mechanisms, technical standard-setting bodies, and mini-lateral cooperation, while cautioning against overstating the likelihood of near-term convergence given divergent political economies and strategic interests.
