How Algorithmic Policing Challenges Fundamental Rights Protection in the EU: Lessons From the United Kingdom

Abstract

Police in the United Kingdom are quickly emerging as global pioneers in the design and use of technologies employed for the purposes of algorithmic policing: i.e., the use of one or more algorithmic tools to generate outputs capable of supporting decisions taken by policing authorities in pursuit of their public duties and obligations to protect the public and/or to detect, prevent and investigate criminal and/or administrative offending. The range of technologies employed for these purposes include live facial recognition tools, geographical ‘hotspot’ mapping tools, and tools which allow predictive risk assessment of an individual’s likelihood of future offending. These technologies are often framed as efficient and effective means by which to overcome such resourcing pressures, whereby digital data is increasingly viewed as a valuable ‘strategic asset’ through which to streamline and improve the precision and pace of policing interventions.

However, despite widespread use of these tools, two significant regulatory challenges remain. Firstly, how to best ensure safeguards for the fundamental rights of individuals, when very little is known about how these tools are designed and intended to operate – even though multiple controversies have proven that their use is a threat to the ongoing protection of fundamental rights? Secondly, how to ensure accountability for the actions of law enforcement authorities, where violations of such rights might have occurred, or may be likely to occur in future: a problem that extends beyond UK, to the operation of algorithmic tools by law enforcement authorities within the EU?

This chapter seeks to evaluate the regulatory landscape of algorithmic policing in Europe, including developments relating to the EU’s AI Act, using examples drawn from the United Kingdom. It begins by defining ‘algorithmic policing’, before presenting and working through key regulatory challenges in this context: i.e., the ‘naming’ and ‘claiming’ fundamental rights. It demonstrates that algorithmic policing tools have a reach beyond law enforcement as such, incorporating wider administrative agencies into policing practices, and vice-versa, whereby tools may be presented as benevolent or beneficial to those being algorithmically profiled and/or evaluated. It ultimately argues that complex webs of administrative relationships developed as part of these programmes, both technical and institutional, further complicate the already-complicated process of ‘naming’ and ‘claiming’ fundamental rights within an algorithmic decision-making process, and that existing regulations – including the AI Act – may not adequately acknowledge this complexity in a way that appropriate safeguards affected persons from harm.

Keywords: Algorithmic Policing, AI Act, Fundamental Rights, Administrative Law, AI Regulation