Risk Scoring and Algorithmic Systems in Law Enforcement: Explainability, Human Oversight and the Right to Challenge

Abstract

This report examines the use of algorithmic risk assessments in law enforcement and the safeguards required when a score influences an individual's treatment. It analyses data quality, the object of prediction, classification errors, discriminatory effects and the distinction between an intermediate assessment and a decision with legal consequences. The comparative framework includes EU data protection law, the AI Act, UK legislation and selected decisions concerning automated assessment, passenger data, facial recognition and risk profiling. Drawing on cases including SCHUFA, Dun & Bradstreet, Bridges and SyRI, the study considers intelligible explanations, meaningful human oversight, documentation and the ability to contest an assessment. It distinguishes technical performance from lawful decision-making and formal human involvement from independent scrutiny. The report's central finding is that explanation, correction and review must operate together: access to a description of the model is insufficient if the individual cannot challenge the data, reasoning or resulting interference.