CEPS op-ed on the AILD

In its 2025 work programme, the European Commission effectively scrapped the AI Liability Directive – a move that threatens to unravel trust in the EU’s burgeoning AI policy landscape. This abrupt decision strips away potential critical protections for victims of AI-related harm and denies businesses the legal certainty they need to innovate.

Find below my op-ed with CEPS (Centre for European Policy Studies) on 'An AI Liability Regulation would complete the EU’s AI strategy' that I wrote for my panel at yesterday's Ideas Lab 2025 with Artur Bogucki, Kazimierz W. Ujazdowski, Ioanna Mazilescu, Dr Alexandra von Westernhagen, and Jan De Bruyne. It features some of Axel Voss and my ideas for the blocked draft report.

To be honest, I needed some time to digest the European Commission’s decision to withdraw the AI Liability Directive (AILD). What for me - a big fan of the EU - was particularly frustrating to see is that this high-level decision rests on three flawed assumptions that would not have withstood any proper discussion:

▪️ "Typical case of EU overregulation": Incorrect. The work on an AI liability law was not coming out of the blue but was from the beginning (2018) a crucial part of the EU AI strategy, based on extensive research in and outside of the EU Institutions. Many political decisions during the AI Act trilogues were taken, assuming that the AILD would later close the remaining gaps.

▪️ "It creates additional red tape": False. The law was designed as an ex-post redress mechanism, ensuring compensation for AI-related harm but without imposing direct compliance burdens for companies. As the majority of products (85%) are insured, concerns over additional costs are also misplaced. In reality, the law would have helped EU companies, particularly those that became victims themselves (i.e. IPR, professionally used property).

▪️ "PLD covers already everything": Misleading. There is robust research (i.e. European Parliamentary Research Service) that clearly demonstrates significant fragmentation and legal uncertainty. It is fair to say that neither the Product Liability Directive nor the national liability frameworks offer an effective solution that tackles the specific legal challenged posed by AI.

While no one was really happy with the Commission's original proposal, there was a large group of people in the European Commission, Council of the European Union, and European Parliament eager to start working on it. A group that was excited to listen to stakeholders, open to rethink previous decisions, and willing to integrate completely new approaches. Our own public consultation was part of this larger 'Better Regulation' campaign for more accessible, transparent, and evidence-based EU policy-making.

Without success: the withdrawal leaves companies and citizens in a legal limbo. And it weakens the overall EU AI strategy, thwarting the envisioned "Ecosystem of Trust". Yesterday's panel at CEPS (Centre for European Policy Studies) indicates nevertheless that although this battle is lost, the journey will continue somehow. Too obvious the necessity that something must happen, too strong the alignment among the EU Institutions and, too obvious the advantages of harmonised AI liability rules. It will just take much longer for the EU to deliver ...