China’s Supreme People’s Court issues first judicial opinions on adjudication of AI-related disputes

5 October 2026

On 7 September 2026, the Supreme People’s Court of the People’s Republic of China (“SPC”) issued the Opinions on Lawfully Adjudicating Cases Involving Artificial Intelligence Disputes (Fa Fa [2026] No. 10) (“Opinions”), marking the first time China’s highest judicial authority has issued a document setting out adjudication rules specifically for AI-related cases.

China does not as yet have a dedicated AI law - the Opinions therefore provide guidance to courts at all levels on applying existing legislation to disputes involving AI. The relevant legislation includes the Civil Code, the Cybersecurity Law, the Data Security Law, the Personal Information Protection Law, the Anti-Unfair Competition Law, and the Consumer Rights Protection Law.

Of particular relevance to businesses deploying AI in China-facing marketing, e-commerce, and customer engagement, the Opinions address the allocation of liability between AI developers, service providers and users, while also identifying consumer-facing practices that may attract closer judicial scrutiny.

This article sets out some of the Opinions’ key highlights.

Key provisions

Determining liability for AI-related harm

The Opinions direct courts to apply the fault-based liability rule in Article 1165(1) of the Civil Code as the general approach to AI-related disputes, except where the law expressly provides for no-fault or presumed-fault liability.

In determining whether a party is at fault, courts will take into account the circumstances of the AI application, including its degree of autonomy, the transparency of the underlying technology and information, the nature and extent of potential risks, and the scope of any resulting impact. The assessment will also consider the preventive measures implemented by AI developers and service providers, the technical feasibility of such measures, and the extent to which users could reasonably foresee and control the harm.

Protection against unauthorised use of personal likeness and voice

The Opinions address the use of AI to reproduce an individual’s identity without consent. Unless otherwise provided by law, where AI is used without a natural person’s consent to process their name, portrait or similar attributes to generate an identifiable virtual digital image, which is then used or published, courts will support the individual’s claim that their name and portrait rights have been infringed. Similarly, where a person’s voice is used without consent as training material to produce an identifiable synthetic voice, courts will support that person’s claim that their voice interests have been infringed.

Where AI-generated images or voices are manipulated to engage in improper conduct or convey false statements that diminish an individual’s social evaluation, reputation rights may also be infringed. The protection extends to deceased persons, with Article 994 of the Civil Code allowing their close relatives to bring claims in respect of unauthorised virtual images.

In addition to determining liability, courts may issue personality rights injunctions requiring infringers to cease the relevant conduct. Such orders may also be directed at network service providers or generative AI service providers, requiring them to discontinue the relevant services to the extent necessary.

Use of publicly disclosed personal information for AI training

The Opinions clarify the circumstances in which personal information that has been lawfully made public may be used for AI model training. Processing information voluntarily disclosed by an individual, or otherwise lawfully disclosed, will generally not constitute an infringement provided that the processing remains within a reasonable scope and the individual has not expressly objected. However, consent remains necessary where the processing significantly affects the individual.

Whether processing falls within a reasonable scope will depend on several considerations, including its necessity and appropriateness in relation to the model’s functions, the nature and sensitivity of the information, its potential impact on the individual, and the original context of disclosure, including how the information could reasonably be expected to be used.

The Opinions separately address the misuse of publicly accessible information for targeted exposure of private details. Using AI to track and analyse a particular individual’s publicly available telephone numbers, user accounts and social media information to obtain and disclose private information, commonly known as “doxxing”, constitutes an infringement of privacy rights.

Responsibility of generative AI service providers for infringing content

Generative AI service providers may incur liability for content that infringes an individual’s reputation, privacy, or other personality rights where they fail to take necessary action following notification by the affected rights holder. Such action may include promptly stopping the generation of infringing content.

To trigger this obligation, the rights holder’s notice must contain preliminary evidence of the alleged infringement and information establishing the rights holder’s real identity.

The Opinions also distinguish between the responsibilities of providers and users. A user who deliberately induces an AI system to generate infringing content, including through the use of infringing prompts, is responsible for their own conduct. However, under Article 1195 of the Civil Code, a service provider may be held jointly liable with the user if it fails to implement the necessary measures after receiving notice of the infringement.

Consumer protection: Personalised pricing and AI-generated endorsements 

The Opinions address certain commercial applications of AI, particularly algorithm-driven pricing and the use of synthetic celebrity endorsements.

Businesses that use algorithms to impose unreasonable differences in pricing or other trading conditions for identical goods or services may incur tort liability where such practices cause damage. In assessing whether differential treatment is unreasonable, courts will consider its effect on consumers’ rights to information, freedom of choice, and fair trading, including whether those rights have been materially restricted or harmed.

Other relevant considerations include whether the trading terms are personalised by reference to factors such as an individual’s consumption preferences, willingness or ability to pay, or browsing history; whether the practice is consistent with good faith and business ethics; and whether the grounds for differential treatment are legitimate, sufficient, and non-discriminatory.

The Opinions also address AI-generated celebrity impersonations used in commercial promotions. Where such impersonation amounts to fraud, affected consumers may seek punitive damages under Article 55 of the Consumer Rights Protection Law.

In light of these provisions, retail businesses should consider reviewing the personalised pricing and promotional mechanisms used across their China-facing e-commerce and membership platforms. They should also maintain records of the relevant authorisations for AI-generated brand ambassadors, virtual models, and synthetic voices featured in their marketing activities.

Evidentiary requirements and disclosure of AI use in court proceedings

The Opinions underscore the importance of maintaining records relating to AI development and use, particularly in the context of copyright disputes.

Where an AI developer relies on a non-infringement defence in copyright litigation, courts will require the developer to produce supporting materials, including information on training data sources, records of the training process, details of the model’s operating mode and the scientific basis for the defence.

The Opinions also recognise the protection of lawfully acquired data rights. Data collections may qualify for protection as trade secrets or, where they do not meet the requirements for trade secret protection, under Article 13 of the Anti-Unfair Competition Law.

Separately, litigation participants using AI to prepare litigation documents or case research reports must verify the accuracy of those materials and disclose their use of AI assistance to the court.

Reference materials

The materials referenced in this article are available on the People’s Republic of China Government websites as set out below.