JUST-AI at ICON·S BeNeLux

The JUST-AI Team will be at ICON·S BeNeLux today to discuss systemic challenges to EU AI Regulation.

The European Union (EU) is facing tremendous challenges in the face of data and digital technologies, chief among them being artificial intelligence (AI). New technology regulation has taken place in a context of political crisis concerning, inter alia, the legitimacy of EU institutions, the erosion of fundamental rights in some Member States, the need for tech-driven economic growth, environmental protection, and the promotion of EU citizens’ socio-economic rights. Designed and enforced in such a context, AI regulation has sought to pursue a dual objective: fostering the technological development of a European AI industry (an ecosystem of excellence), while simultaneously ensuring that AI technologies do not adversely affect the protection of fundamental rights (an ecosystem of trust). The EU legislature’s response to this balancing undertaking is twofold. Substantively, the EU adopted Regulation 2024/1689 laying down harmonised rules on AI (AI Act - AIA). Procedurally, the EU revised the Product Liability Directive. Interestingly, this instrument was meant to be completed by a new directive on non-contractual civil liability rules to AI, submitted as a proposal in 2022, but withdrawn in February 2025.

Against that backdrop, Noémi Bontridder’s contribution will lay out the broader context in which EU AI regulatory policy has emerged and examine the tensions between precaution, innovation, and uncertainty. Building on this, Jerome De Cooman will argue that the AIA represents a normative revolution in the EU’s broader legislative landscape, as it applies a product safety logic, intended to promote both innovation and precaution (by promoting safety
standards). The AIA achieves these goals through a combination of pre-market (self) certification, and post-market monitoring (by both providers and supervisory authorities). The end goal is that such a regulatory design be conducive to sufficient protection of EU citizens against fundamental rights infringements associated with the so-called high-risk AI systems. It is precisely the role of fundamental rights within this regulation that Anca Radu will seek to question. In essence, her contribution examines the relevance of standardising the protection of fundamental rights. She will address this issue through the vantage point of the right of individuals to obtain human intervention (or explanation) in connection with automated (and potentially opaque and biased) decision-making. To complete the picture, Ljupcho Grozdanovski will analyse the effectiveness of procedural fairness in the context of AI liability. If an AI system causes harm, proving that harm and how it causally relates to the system remains challenging. The AIA contains no evidentiary framework and the PLD’s scope remains limited to – in essence – material harm. Ljupcho Grozdanovski will thus analyse if the absence of adequate procedural frameworks may compromise the AIA’s private enforcement.