China’s Supreme People’s Court has issued the country’s first judicial policy document establishing rules for handling disputes involving artificial intelligence, Qazinform News Agency reports.
The Opinion on Lawfully Adjudicating Cases Involving Artificial Intelligence Disputes, published on September 7, consists of five sections and 24 provisions. It covers deepfakes, AI-generated content, algorithmic pricing, personal data, autonomous vehicles, intellectual property and the use of AI in court proceedings.
The court described it as the first document of its kind issued by China’s highest judicial authority. However, it is not a separate AI law. The Opinion provides guidance to courts on applying existing legislation, including the Civil Code, Cybersecurity Law, Data Security Law, Copyright Law, Personal Information Protection Law and Consumer Rights Protection Law, to disputes arising from AI technologies.
China has not yet adopted a comprehensive law specifically regulating artificial intelligence, the court noted.
Among the document’s key provisions are:
Liability for AI-related harm. Unless existing legislation provides for strict liability or a presumption of fault, courts should apply the fault-based liability principle. In determining responsibility, judges must consider the AI system’s level of autonomy and transparency, the risks involved, measures taken by developers and providers, and the user’s ability to foresee and control the potential harm.
Deepfakes and voice cloning. Creating, using or publishing an identifiable digital representation of a person without their consent may constitute an infringement of their rights to their name and likeness. Using a person’s voice as training data to imitate their tone, pronunciation and speaking style without consent may also violate their rights. Manipulating a digital image or synthetic voice to spread false statements that damage someone’s reputation may result in legal liability.
Digital recreation of deceased people. Unauthorized creation or use of an AI-generated representation of a deceased person may lead to civil liability if it harms the person’s name, likeness, reputation or other protected interests. Close relatives may bring a claim under the Civil Code.
Privacy and doxxing. Using AI to track and analyze telephone numbers, online accounts and publicly available social media information in order to obtain or disclose private information may be recognized as an infringement of privacy.
Personal data used for AI training. Processing lawfully disclosed personal information within reasonable limits for model training will generally not be considered an infringement if the individual has not explicitly objected. However, consent must be obtained when the processing could have a significant impact on the person’s rights and interests.
Responsibility of generative AI providers. If AI-generated content violates a person’s reputation, privacy or other personal rights, a service provider may be held liable if it fails to take necessary action after receiving a valid notice. A user who deliberately prompts an AI system to generate harmful or unlawful content will also bear responsibility.
Algorithmic price discrimination. Businesses may be held liable if they use algorithms to offer the same product or service to different consumers on unreasonably different terms. Courts should assess whether the practice restricts consumers’ rights to information, independent choice and fair transactions, and whether individual prices are based on browsing history, willingness to pay or purchasing power.
AI-generated celebrity endorsements. Consumers may seek punitive damages if a company uses AI to impersonate a celebrity while selling products or services and the practice amounts to fraud.
Autonomous and assisted-driving vehicles. Compensation following an accident must be determined under the Civil Code and road traffic legislation. Manufacturers or sellers may be held responsible when a vehicle defect causes an accident. If the damage results from both a defect and the driver’s actions, claims may be brought against both the driver and the manufacturer or seller. Courts may also require manufacturers, sellers or operators to provide driving-system records needed to establish the cause of an accident.
AI-generated content and copyright. Responsibility for copyright infringement must be allocated among developers, service providers and users based on factors including the type of service, source of training data, each party’s involvement, preventive measures and financial benefit. A user may be held liable if they knew or should have known about an earlier work and used AI to create substantially similar content without a valid legal defense.
The court has not yet taken a position on whether content generated by AI itself qualifies for copyright protection or whether using copyrighted works without permission to train large AI models constitutes infringement. Officials said these issues remain disputed and require further study.
The Opinion also provides possible liability exemptions for developers and providers of open-source software that make code modules available free of charge and clearly disclose their functions and security risks.
For AI-related inventions, a natural person may be recognized as the inventor only if they made a creative contribution to the invention’s essential features. The rules also address the use of artificial intelligence in judicial proceedings. Parties must verify the accuracy of AI-generated legal documents, case research reports and other materials before submitting them to court. They must also disclose the use of AI and remain responsible for the authenticity and accuracy of the information provided.
Using artificial intelligence to fabricate evidence or initiate fraudulent litigation may result in the dismissal of claims, fines, detention or criminal prosecution. Earlier, Qazinform News Agency reported that California Governor Gavin Newsom had signed an executive order directing the development of new rules governing artificial intelligence.
The agency also reported on Kazakhstan’s plans to become one of the region’s leaders in artificial intelligence. The country declared 2026 the Year of Digitalization and Artificial Intelligence, established a dedicated ministry, adopted an AI law, and launched a series of educational and infrastructure projects.