Artificial intelligence in law enforcement: a critical analysis of the EU AI Act’s framework for protecting public values and fundamental rights

The increasing deployment of artificial intelligence (hereinafter “AI”) systems by law enforcement authorities across the European Union has intensified long-standing concerns regarding the protection of fundamental rights, the rule of law, transparency, and democratic accountability. While AI-based technologies promise enhanced efficiency, predictive capacity, and resource optimisation in policing, their use in this particularly sensitive domain risks exacerbating discrimination, opacity, and arbitrary decision-making. These risks directly implicate the core values enshrined in Article 2 TEU and demand a robust and coherent regulatory response. Against this backdrop, the present Thesis examines whether the existing European legal framework, most notably Regulation (EU) 2024/1689, is adequate to ensure that AI systems deployed by law enforcement authorities operate lawfully, fairly, and in full compliance with fundamental rights and public values. To that end, the analysis critically engages with the AI Act’s regulatory architecture as applied to two of the most contested uses of AI in policing: predictive policing and automated video surveillance. To address the research question, the Thesis investigates three interrelated issues: first, whether the AI Act’s definition of “AI system” adequately captures law enforcement systems that pose significant risks to fundamental rights; second, whether the Regulation’s concept of “law enforcement” is sufficiently clear to ensure legal certainty and consistent risk classification; and third, whether the AI Act, in conjunction with EU data protection law, provides substantive protection beyond minimum standards, or instead leaves structural gaps in the governance of algorithmic policing. Methodologically, the Thesis adopts a doctrinal legal analysis, complemented by interdisciplinary insights from computer science, criminology, and ethics, which are necessary to understand both the technological object of regulation and the policing strategies that inform its use. The Thesis argues that while the AI Act represents a significant step forward in the EU’s approach to AI governance, its application to law enforcement remains marked by conceptual ambiguities, regulatory fragmentation, and enforcement challenges. These shortcomings risk undermining the effective protection of fundamental rights and public values unless addressed through careful interpretation, complementary safeguards, and a principled application of transparency and fairness requirements tailored to the law enforcement context.