AI Disputes in China: New Guidelines from the Supreme People’s Court

China’s Supreme People’s Court publishes opinion on the adjudication of AI disputes

Without prior notice, China’s Supreme People’s Court published the “Opinion of the Supreme People’s Court on the Lawful Handling of Disputes Relating to Artificial Intelligence” (hereinafter referred to as the “Opinion”) on 7 September 2026. The Opinion comprises five parts with 24 articles and represents the first comprehensive judicial guideline from China’s highest court on civil and criminal AI disputes.

The Opinion is not a formal law. Unlike the EU AI Regulation, China does not currently have a unified “AI law”. The Opinion fills this gap by setting out how courts should apply existing provisions of the Civil Code, the Copyright Law, the Competition Law, the Data Protection Law, the Product Quality Law and the Criminal Law to AI-related cases.

Although not technically binding, the Opinion has significant practical impact, as lower courts follow the Supreme People’s Court’s interpretations to ensure consistent case law.

The indirect core message of the Opinion is that AI systems are tools like any others and must be treated as such.

Key provisions

Courts are to take into account the usage scenario, the degree of autonomy of the system, transparency, potential risks, and the protective measures taken by developers, providers and users. In the absence of a legal basis for no-fault liability, the general provisions of tort law shall generally apply.

Rights of personality

With regard to rights of personality, the Opinion makes it clear that the creation of recognisable digital avatars or synthetic voices without the consent of the data subject constitutes a legal infringement; if such content is used to make false claims, this also constitutes a violation of the right to honour. At the same time, the processing of lawfully published personal data for training purposes, to a reasonable extent, is not generally considered unlawful, provided the data subject has not objected. Providers of generative AI services who fail to act in a timely manner following a report of an infringement are liable under applicable law.

Consumer protection

In the context of consumer protection, the imitation of celebrities by AI for fraudulent sales campaigns is deemed to be fraud, giving rise to a claim for punitive damages under the Consumer Protection Law. Algorithmic price discrimination (“big data killing”) gives rise to liability for breach of fair trading conditions.

In the event of damage caused by faulty AI-integrated devices, the existing product liability provisions apply unchanged; in the event of accidents involving driver assistance or autonomous systems, liability is apportioned on the basis of actual causality, taking into account vehicle defects and driver fault, whereby courts may order the submission of necessary data.

Intellectual Property

With regard to intellectual property rights, liability for AI content that infringes copyright is apportioned amongst developers, providers and users according to their role, control and contribution; defendants may be required to disclose training data. The use of AI for product piracy or misleading advertising falls under competition law. For AI-assisted inventions, a named human inventor remains necessary.

With regard to evidence, the Opinion calls for a comprehensive examination of the authenticity of AI-generated evidence, for example by examining the prompts used and assessing reproducibility. Parties submitting AI-generated pleadings must verify their accuracy and disclose the use of AI. Breaches may lead to adverse assessments of evidence or sanctions. Abusive falsification of evidence using AI may result in the dismissal of claims, fines, imprisonment or criminal prosecution. From a criminal law perspective, the Supreme Court considers the existing Criminal Law to be sufficient to cover AI-assisted fraud, defamation and data offences.

According to statements made at a press conference held by the Supreme People’s Court on 7 September 2026 to mark the publication of the Opinion, two key questions remain unresolved due to significant differences of opinion: whether content generated autonomously by AI is eligible for copyright protection, and whether the use of protected works for training purposes without consent constitutes an infringement. Both issues are to be left to future case law or legislation.

Conclusion

The Supreme People’s Court’s new guideline establishes a consistent framework for the judicial handling of AI-related disputes in China for the first time. Although it does not constitute a formal law, it will have a significant impact on the application of the law in practice. Companies should take these requirements into account at an early stage, particularly in the areas of liability, data protection, intellectual property and the documentation of AI applications.

The general guideline forms the overarching legal framework. The following articles on the guidelines issued by the Shanghai Bar Association and the NFRA guidelines for financial institutions illustrate how individual sectors are implementing these requirements in practice.

• Part 2: Practical Guidelines of the Shanghai Bar Association
• Part 3: AI Compliance in the Financial Sector
• Go to the overview page for the series “New AI Regulation in China 2026”