Algorithmic Governance and Legal Accountability: A Comparative Analysis of Artificial Intelligence Regulation in Public Decision-Making

Abstract

This article examines the problem of legal accountability in algorithmic governance, particularly in the context of artificial intelligence (AI) used in public decision-making, where the increasing reliance on automated systems generates significant challenges related to transparency, liability attribution, and the protection of fundamental rights. The study addresses the normative and practical gaps arising from the coexistence of binding regulations, such as the European Union Artificial Intelligence Act, and non-binding frameworks, including the OECD AI Principles and UNESCO Recommendation, alongside fragmented national policies. Employing a normative-comparative legal research method, this study utilizes doctrinal analysis and comparative approaches based on primary legal sources and secondary scholarly literature to evaluate the coherence, consistency, and effectiveness of AI regulatory regimes across jurisdictions. The findings reveal that while the EU framework provides a relatively structured and enforceable model of accountability through a risk-based regulatory approach, significant ambiguities persist in the allocation of legal liability and the operationalization of human oversight, particularly when algorithmic systems influence administrative discretion. Furthermore, the analysis demonstrates that socio-legal factors, including institutional capacity, legal culture, and administrative practices, critically affect the implementation of accountability norms, thereby exposing a gap between formal legal design and empirical realities, especially in developing regulatory environments. The study contributes to legal scholarship by proposing an integrated framework of algorithmic accountability that combines normative clarity, comparative insights, and socio-legal considerations to strengthen governance mechanisms in AI-driven public administration