From Accountability to Liability: How Accountability Mechanisms Shape the Non-Contractual Liability of Generative AI in the EU and China

The article compares how accountability mechanisms shape non-contractual liability for generative AI in the European Union and China. Through a structured analysis of each regime’s approach, it identifies two distinct models: the EU’s horizontal, ex ante framework and China’s vertical, sector-specific framework. The study finds that accountability measures lead to liability outcomes only when they are legally formalised and provide enforceable standards or evidence for courts. In the EU, mandatory risk controls, transparency, and record-keeping create clear benchmarks and an auditable trail that facilitate fault attribution under updated liability laws. In China, accountability remains largely an administrative oversight tool; compliance records do not automatically translate into liability standards, making any civil redress contingent on prior regulatory action. The article concludes that accountability is a necessary precursor to effective AI liability, but it must be designed as a legally actionable and judiciary-ready mechanism to bridge the “accountability gap” in AI governance.