Hong Wu (Faculty of Law, University of Oxford; NingboTech University – School of Law) has posted Special Tort Liability Regimes for Harm Arising from Artificial Intelligence in China on SSRN. Here is the abstract:
Determining and assigning tort liability for artificial intelligence (AI) represents one of the most complex and critical challenges in AI legislation. The traditional product liability framework in China is inadequate to address three core issues arising from AI-related torts: the difficulties of proving fault and causation, defining new types of harm, and identifying liable parties. A more effective approach would build upon the foundation of fault-based liability, while introducing targeted provisions to ease the evidentiary burden on claimants. Victims can only overcome the evidentiary asymmetry inherent in AI systems by gaining access to development documentation and system activity logs. Accordingly, legislation should establish rules for evidence disclosure and impose conditional information disclosure obligations on relevant AI stakeholders, thereby providing a substantive legal basis for courts to order the production of documents. In the AI era, legal remedies for virtual harm should not rely on an expanded definition of material damage. Instead, the current requirement of “seriousness” for emotional distress compensation should be replaced with a “significance” standard. To alleviate the burden of proving causation for consumers of AI products, the law should introduce presumptions of causality under defined circumstances. Where harm clearly results from a breach of duty but the precise liable actor cannot be identified, all entities within the same commercial and technical unit should bear joint and several liability.