A Typology of Automated Law Enforcement Systems Under EU Law: What Standards for Defence Rights?

New and emerging technologies are re-shaping police-citizen encounters – ranging from person-based predictive policing software deployed to spot likely wrongdoers to AI-driven software solutions employed to collect, organise, visualise and assess data for criminal investigation purposes. Hence, automated law enforcement requires a careful balancing act between effectiveness and innovation on the one hand, and protection of fundamental rights on the other. It is no coincidence that the EU AI Act classifies AI systems of this or similar kind in principle as high-risk systems and establishes functional requirements and additional safeguards in the form of legal obligations on the providers and deployers thereof. Additionally, the 2021 European Parliament's Resolution on AI in criminal law further specifies the fundamental rights scrutiny required in the case of automated law enforcement. Against this background, this article first presents a typology of automated law enforcement systems and discusses the overall approach in the EU AI Act towards the use of AI in the area of criminal law. Next, it focuses on defence rights standards, exploring how these standards may be reshaped at the EU level, incorporated into technical design, and enforced nationally. Such a combined solution should ensure that defence rights standards are adapted to the potential of the machine and to the power imbalance and fallibility inherent in its use.