Another expert study, same conclusion: withdrawing the AI Liability Directive would be a serious strategic mistake

A new study by Prof. Andrea Bertolini confirms what many experts have long argued: scrapping the AI Liability Directive (AILD) would be a strategic error for the EU. The analysis shows that the revised Product Liability Directive (PLD) cannot address the complexities of AI-related harm alone. Instead, a dedicated, harmonised liability regime is needed to protect victims, reduce legal uncertainty, and ensure fairness across the AI value chain. Fragmentation is already occurring, and without action, Europe risks undermining its innovation and competitiveness. How much more evidence do EU leaders need to finally act?

In his latest analysis for the European Parliament, Prof. Andrea Bertolini dismantles the idea that the updated Product Liability Directive (PLD) alone can handle the complexity of AI-related harm. He argues persuasively for a standalone, strict liability regime targeting high-risk AI systems ... exactly the kind of harmonised approach Axel Voss and I have been calling for six years now.

The study echoes our key concerns:

▪︎ National fragmentation is no longer a risk, it is already happening.

▪︎ The PLD is limited in scope and ineffective in many AI scenarios.

▪︎ An improved AILD proposal - if done right - would protect victims of AI, lower litigation costs, and reduce legal uncertainty; especially for Europe's SMEs and start-ups.

▪︎ Full alignment with the AI Act and clear liability along the AI value chain are essential to ensure fairness and innovation.

This brings me to the question: how much more scientific evidence do EU leaders need before they admit it? This is the second study of an independent researcher for the Legal committee (JURI) on AI liability in less than 12 months and one of many for the EU Institutions over the past 10 years. They all indicate the same: scrapping the AILD would not simplify the legal landscape but would harm our competitiveness (Draghi⚡️) and entrench chaos among the victims of AI such as citizens and EU companies.

❓Isn‘t it time to face the facts and fix this before it is too late❓

Download the study here:

2025.07.24 AILD study.pdf

Download PDF • 3.21MB