Governing Artificial Intelligence: Rule of Law, Democratic Accountability and the Politics of Regulation in a Fragmented World

Abstract

Artificial intelligence has moved from a technical curiosity to a central force in economic, social and political life. As governments race to regulate it, a patchwork of laws and voluntary codes has emerged, led by the European Union's AI Act, the United States' mixture of federal caution and state-level activism, and Asian approaches ranging from Singapore's agentic AI framework to China's algorithm-specific rules. This review paper uses a socio-legal method to examine whether these frameworks genuinely protect the rule of law, democratic accountability and fundamental rights, or whether they mainly manage economic competition between trading blocs. The paper argues that current AI governance suffers from three connected problems: a widening gap between binding law and fast-moving technology, a concentration of regulatory capacity in a small number of wealthy states and firms, and a weak link between formal legal compliance and genuine democratic control over automated power. Drawing on recent regulatory developments up to mid-2026, including the EU's Digital Omnibus and the rise of autonomous AI agents, the paper offers a critical assessment of enforcement gaps and proposes a rights-centred, participatory model of governance that treats AI regulation as a question of political power, not only of technical risk management.