Judicial Oversight in the Age of AI: Can the CJEU Effectively Review Automated Risk Assessments?

Abstract

The European Union’s security architecture is undergoing a profound algorithmic turn, transitioning from reactive law enforcement to preemptive, risk-based border control. Instruments like the European Travel Information and Authorisation System (ETIAS) and the upgraded Europol mandate increasingly rely on Automated Risk Assessments (ARA) and Machine Learning to flag high-risk individuals before a crime is committed. This article critically examines whether the Court of Justice of the European Union (CJEU) possesses the epistemic competence to effectively review these opaque, probabilistic determinations. The paper argues that the CJEU’s traditional administrative law standards, specifically the deferential manifest error of assessment test and the human in the loop safeguard, are structurally inadequate for the age of AI. The technical opacity of Black Box algorithms, combined with the proprietary nature of risk indicators, creates a blind spot in judicial protection that threatens the essence of Article 47 of the Charter of Fundamental Rights. By analyzing recent jurisprudence and the emerging AI Act, this article proposes a normative shift from a substantive review of outcomes to a procedural review of algorithmic design. It advocates for a new techno-judicial standard of scrutiny, necessitating the reversal of the burden of proof, strict explainability mandates for high-stakes adjudication, and the appointment of court-mandated technical experts. Ultimately, the article posits that without these procedural innovations, the CJEU risks becoming a rubber stamp for a Technological Leviathan, leaving the fundamental rights of data subjects without an effective remedy.