AI and Liability: Who is responsible for Mistakes made by Artificial Intelligence?

Feat. Carsten Gerner-Beuerle of University College London | 13 February 2024

The risks associated with the use of AI are widely discussed. Examples range from possible misdiagnoses in medical treatment and discrimination in job searches to dangers for democratic systems or violations of human dignity. The possibly existential threats in particular, which are said to emanate from generative AI and which its creators themselves never tire of pointing out, are a source of heated debate. However, AI systems do not emerge and act on their own: Humans are always involved in their development, dissemination and use, at least for the time being, and humans still bear responsibility for how the systems are set up, what they do and what happens to the results. But are they also liable when mistakes happen? It is not easy to decide who is held responsible if an AI system causes damage, and our legal situation does not yet have any clear answers to this question. This is about to change soon: With the AI Act, the EU is currently in the process of establishing product liability for AI. What are the aspects which lawyers and legislators need to address when it comes to liability issues involving the use of AI? And how well is this upcoming AI law compatible or at odds with our existing understanding of the law?

Carsten Gerner-Beuerle is Professor of Commercial Law at University College London. The Law and Economics scholar is an expert in International Corporate, Private and Capital Market Law and has contributed to/weighed in on regulatory policies of the EU Commission and the EU Parliament. In this episode of Digitalgespräch, he describes the unclear liability situation in which AI is used so far and how it affects current practice. He explains the legal background to this complex situation and describes how different legal systems deal with risks arising from the use of AI. Furthermore, he outlines the strategy the EU is pursuing with the AI Act. Together with hosts Marlene Görger and Petra Gehring, Gerner-Beuerle discusses the characteristics of AI that distinguish it from “ordinary” products and how liability is usually addressed, the typical risks associated with the use of AI systems in general and the extent to which the concept of liability could nonetheless function as a regulatory instrument.