The webinar presented and discussed ELI’s 1st Supplement to the ELI Draft of a Revised Product Liability Directive, examining whether providers of social media platforms, AI chatbots and comparable digital consumer offerings may be liable under the EU’s revised Product Liability Directive (PLD) for harms caused by their products.
Recording below.
The webinar was moderated by Sir Geoffrey Vos (ELI First Vice-President; Master of the Rolls and Head of Civil Justice in England and Wales) and featured presentations by Christian Twigg-Flesner (Professor, University of Warwick), Christiane Wendehorst (ELI Scientific Director; Professor, University of Vienna; former ELI President) and Pascal Pichonnaz (Professor, University of Fribourg; former ELI President).
Opening the webinar, Sir Geoffrey described product liability and digital harms as a high-profile and increasingly important area at the intersection of private law, EU law and emerging litigation concerning social media, AI and other digital services.
Christian Twigg-Flesner traced the development of ELI’s work on product liability, beginning with its 2021 Innovation Paper on Guiding Principles for Updating the Product Liability Directive for the Digital Age and continuing through its contributions to the European Commission’s public consultation (available here and here) and its Draft of a Revised Product Liability Directive. He explained that ELI had consistently advocated extending the product liability regime to digital products, content and services, while adapting concepts such as defectiveness, damage and the placing of products on the market to the realities of the digital economy.
Many of ELI’s recommendations were reflected in the revised PLD adopted in 2024. However, the Directive’s decision to refer to ‘software’, rather than the established concepts of digital content and digital services, leaves important questions concerning its application to social media platforms, AI chatbots and comparable offerings.
Presenting the new ELI Supplement, Christiane Wendehorst explained that it does not seek to prescribe a definitive interpretation of the Directive, which will ultimately be a matter for the courts. Instead, it provides a practical legal framework by identifying three possible routes to liability: treating the digital offering itself as software placed on the market; treating the provider as a manufacturer of software developed for its own use; or treating the platform service as a related service connected with an app or other software component.
The Supplement also proposes concrete criteria for distinguishing between software used by consumers from a service provided through software. These include who chooses to use the software, who determines when and how frequently it is activated, who supplies prompts or other input, who exercises oversight and who first assesses its output. Applied to companion chatbots, for example, these criteria may support their classification as products used directly by consumers.
Pascal Pichonnaz examined whether the revised regime strikes an appropriate balance between consumer protection, innovation and legal certainty. He focused particularly on the Directive’s rules on disclosure of evidence, presumptions of defectiveness and causation, and the continuing role of national law. While these mechanisms may assist claimants facing technically complex products, significant questions remain concerning the proof of psychological harm, addictive design and the interaction of several possible causes of damage.
The subsequent discussion addressed the distinction between software and information, the position of platforms under the Digital Services Act, fraud committed using AI tools, compulsory insurance, open-source software and the likely development of litigation in Europe. The speakers distinguished between platforms that merely display information and algorithms that actively select, prioritise or repeatedly present content in ways that may cause or reinforce harm.
Looking ahead, the panellists anticipated that questions concerning social media platforms and AI chatbots would reach national courts and the Court of Justice of the European Union. Wendehorst concluded that product liability ‘provides a sound basis’ for addressing digital harms and highlighted ELI’s role in explaining the legal difficulties and assisting courts, practitioners and policymakers in applying the new regime.
The recording is available below.
