Artificial Intelligence & Civil Liability: An economic and comparative legal approach to U.S. & E.U. law

Research project

Investigator: Stefan Heiss

Today, Artificial Intelligence (AI) is already deeply embedded into our social fabric and its effects are unlike anything that have come before. Use cases have outlined that poor-quality AI is operating in public. Its increasing ubiquity will put greater strain on the liability system. The question, thus, becomes: What legislative changes should be made? Since insufficient liability rules might lead the development of the technology into an inadequate direction, further assessments of U.S. and E.U. (focusing also on German) liability rules are needed. When comparing the paths the E.U. and the U.S. are taking on AI, different legislative courses can be discerned: The E.U. is initiating significant adjustments to turn Europe into the global hub for socially desirable AI, while the U.S. have approached piecemeal legislation and recommendations which are more hesitant to subject the new technology to regulation. This project aims to shine a comparative light on these issues and provide an in-depth analysis of existing liability rules and new proposals and thereby seeks to focus on the following aspects: (i) The E.U. proposals pursue a risk-based approach. However, clear-cut parameters are required to distinguish between the riskiness of AI systems. Lessons might be learned from U.S. court decisions on the basic choice between the rule of negligence and strict liability. (ii) Further, doctrinal errors for assigning the major liability rules to AI need a closer analysis. Scholars have indicated that liability rules may work well for humans, but the applicability of tort law doctrines in the context of AI cannot be ascertained in a generalized way. (iii) Due to multi-causal and reciprocal hazards by AI, an evaluation of the burden of proof and the reimbursable losses seems necessary, thereby comparing U.S. and E.U. methods. (iv) In order to close some of the emerging gaps, a scheme of “strict liability to the state” is outlined.Picking the most valuable insights of the U.S. and E.U. approaches on AI may lead to a desirable path. By means of economic analysis and comparative studies of the law, the purpose of the research project is to determine a roadmap for society in terms of civil liability and AI.