This project examines how liability for patient injuries should be distributed along the medical AI value chain. It starts from the rapid uptake of AI in healthcare and the fact that many hospital applications are purchased from external vendors, while some — especially in university hospitals — are developed in-house. The core question is how European tort law, and rules governing liability for patient injuries in different European countries address AI-induced system failure as a deviation from good medical practice and how responsibility should be allocated between producers, hospitals, and users (e.g. clinicians). The project will include analyses from Austria, Italy, France, Norway, Canada and the USA, combining EU jurisdictions with differing liability regimes and countries at the forefront of AI development and testing. Developers and hospital decision-makers will be involved to ensure that the legal analysis reflects technical developments and practical implementation challenges.
ECTIL is a research partner of TRUST (The Norwegian Centre for Trustworthy AI) in this project. The participants will convene for comparative national studies and regular workshops, which will form the basis of authoritative output contributing to the highly active research environment in this field. The project is co-led by Ernst Karner (University of Vienna) and Anne Marie Frøseth (University of Bergen).