AI Security Regulatory Gaps: A Global Comparative Review and a Governance Blueprint Tailored for the United Arab Emirates

Abstract

Artificial intelligence (AI) is being adopted at unprecedented speed across public and private sectors, while regulatory responses diverge widely across jurisdictions. The European Union (EU) AI Act inaugurates a harmonized, risk-based regime; the United States and United Kingdom pursue sectoral or principles led approaches; Singapore and China implement prescriptive models; and Gulf Cooperation Council (GCC) states, including the United Arab Emirates (UAE)have favored nonbinding strategies and sandboxes. Building on a scoping of 48 publications, this review synthesizes advances in AI governance, highlights unresolved security and enforcement gaps, and proposes a UAE tailored framework aligned with ISO/ISMS and international best practices. We find persistent weaknesses in risk classification fidelity, security-by-design obligations, policy–practice enforcement, and governance for general purpose and agentic AI. We propose (i) a global AI security and governance framework and (ii) a UAE governance blueprint with tiered obligations for critical sectors, conformity assessment, incident reporting, and independent audits. The review contributes a practical compliance checklist and legislative recommendations forming the basis for a UAE AI Act, alongside an implementation roadmap.