ABSTRACT
EU law now recognizes artificial intelligence (AI) as a product that can be governed under product safety regulations and product liability rules. AI’s nature as an intangible, dynamic product that is capable of exhibiting varying levels of autonomy, opaqueness, and human-like reasoning make it difficult to assess its defectiveness under the new Product Liability Directive (PLD). Although AI may exhibit human-like behavior, the tort liability system currently does not have a mechanism applying behavioral standards, like negligence, to products. Instead, the PLD’s strict product liability regime seeks to determine whether a product’s condition is defective. This paper argues that AI product defects can stem from the quality of the manufacturer’s risk management activities in connection with the AI. It demonstrates that a ‘risk management defect’ approach is supported by safety regulations governing AI as a product and can help overcome challenges that AI presents to the concept of product defectiveness.
Duffourc, Mindy, Risk Management Rising: Towards a Risk Management Defect Framework for Fair Governance of Black-Box AI Liability. Posted to SSRN 10 May 2025.
Leave a Reply