Diving into the Iceberg: Establishing Transparency in AI for Law Enforcement

The use of AI in security contexts demands rigorous scrutiny and accountability, intersecting multiple layers of regulation. While current discussions on AI transparency often focus on personal data protection and the Artificial Intelligence Act (AI Act), this Article explores how the AI Act, and the Law Enforcement Directive (LED, 2016/680) can be read together to enhance transparency. Using case studies such as Clearview AI, SyRi, and BriefCam, the Article argues that when considered in parallel, these frameworks can offer crucial safeguards against the individual and societal risks posed by AI tools. However, the structural limitations of these regulations, coupled with the practices of key actors, create challenges that may impede their full implementation. By critically analysing both the legislative frameworks and practical examples, the Article demonstrates that, in the absence of insider disclosures and public revelations, these regulations risk failing to enable meaningful public scrutiny, especially in sensitive contexts like national security and predictive policing. The Article identifies significant gaps, including the need for best practices in LED implementation, harmonized regulations across security stakeholders, and most importantly the pressing need for robust rights to information and access, especially when fair administration and fair trials are at stake.