To Örebro University

The draft European Union artificial intelligence (AI) act, unveiled in April 2021, aims to limit the use of biometric identification systems including facial recognition used for surveillance. This proposed legislation focuses primarily on strengthening rules around data quality, transparency, human oversight and accountability. The draft also proposes new rules governing the use of facial recognition technology (FRTs). In June 2023, European Union lawmakers agreed changes to the draft AI act, to include a ban on the use of the technology in biometric surveillance. The draft act is structured on a differentiation system ranging from minimal or no risk, over limited risk and high risk, to unacceptable risk. The latter gives ground for prohibited means, albeit with certain exceptions. Prohibited means could for example be the use of real-time remote FTR systems in the public domain, under peacetime conditions.

The proposed draft act may have implications for military in security operations (MiSO). Although law enforcement in normal conditions is carried out by law enforcement agencies, armed forces may be called on to support civil authorities in situations where the threat level is elevated, the degree of violence or other operational challenges exceeds the capabilities of the traditional law enforcement agencies. Similarly, deploying MiSO may occur in the context of international armed conflict (including occupation) and non-international armed conflict (NIAC). For these situations, the permissive use of biometrics is regulated by an international legal framework that combines human rights law (IHRL) and the law of armed conflict (LOAC). 

This paper examines what implications the draft AI act will have for military security operations (MiSO), with regards to the use of biometrics in law enforcement operations in normal times, through crisis and emergencies, even crossing the threshold of NIAC. It does so by first outlining in what situations use of biometrics may become an issue in MiSO with law enforcement tasks, examining surveillance, arrest and detention, search and seizure, public order management and border management. Next, it examines the potential implications the draft AI act, may have on the use of biometrics in combination with the existing IHRL/LOAC framework. Third, the paper discusses to what extent the proposed risk-based approach aligns with, complements or contrasts with the approach of the IHRL/LOAC regime, both inherently preventive by nature. Specific attention is given to how the notion of risk is addressed in relation to the use of biometrics in these three legal regimes. Lastly, remaining challenges are pointed out and conclusions are drawn.