On September 7, 2026, the Supreme People’s Court of China officially released the Opinions on the Lawful Adjudication of Artificial Intelligence-Related Disputes (Fa Fa [2026] No. 10, hereinafter referred to as the “Opinions”)¹ . As the first judicial adjudication rules document concerning artificial intelligence issued by the country’s highest judicial authority, the release of the Opinions marks a milestone in China’s legal response to the rapid development of AI technology.

In the face of emerging disputes brought about by the flourishing development of AI—such as “AI face-swapping and voice-cloning,” “doxxing,” and “AI resurrection of the deceased”—the Opinions directly address widespread societal concerns and provide clear substantive adjudication standards and procedural rules for people’s courts at all levels to fairly handle AI-related disputes. This not only signifies a crucial step forward in China’s judicial protection framework for artificial intelligence but also lays a solid rule-of-law foundation for the high-quality and sustainable development of the AI industry.

Core Content and Key Points

The Opinions consist of five parts and 24 articles, systematically constructing a comprehensive adjudication framework for AI-related disputes. The core content can be summarized across five major areas:

1. General Principles(Articles 1 – 2)

The Opinions establish the guiding ideology for adjudicating AI-related disputes, centered on three fundamental principles: people-centeredness, support for innovation-driven development, and safeguarding the bottom line of security. The document emphasizes that development and security must be given equal weight, and that promoting innovation must be combined with law-based governance, ensuring that judicial decisions maintain the correct political direction and effectively implement the governance philosophy of balancing development with security.

2. Adjudication of AI-Related Tort Cases (Articles 3 – 11)

The Opinions provide detailed rules for the judicial protection of civil rights and interests in response to socially prominent tort issues:

• Liability Principles: Where AI is used to infringe upon civil rights and interests, the fault liability principle shall generally apply, unless otherwise provided by law. Determining fault requires comprehensive consideration of the specific application scenario, degree of autonomy, technical transparency, potential risks, and the control capabilities of all parties involved.
• Regulating “AI Face-Swapping/Voice-Cloning” and “AI Resurrection of the Deceased”: Using AI to process another person’s name, likeness, or voice without consent to generate identifiable virtual digital avatars or synthetic voices for public use constitutes infringement of the right to name, portrait rights, and voice-related interests. Manipulating virtual avatars to make false statements that lower a person’s social evaluation constitutes infringement of reputation rights. Unauthorized creation or use of a deceased person’s virtual digital avatar shall support close relatives’ claims for civil liability in accordance with the law.
• Regulating “Doxxing” and “Human Flesh Search”: Using AI to track and analyze publicly available information for the purpose of prying into privacy and obtain and disclose private information, or using AI to photograph or eavesdrop on private spaces and activities, shall be deemed an infringement of privacy rights.
• Personal Information Processing in Model Training: Processing lawfully disclosed personal information within a reasonable scope for model training, where the individual has not explicitly refused, shall generally not be deemed infringement. However, where such processing has a significant impact on individual rights and interests, consent must be obtained in accordance with the law.
• Liability of Generative AI Service Providers: The “notice-and-necessary-measures” rule applies. If a service provider fails to take necessary measures—such as stopping the generation of infringing content—after being notified by the rights holder, it shall bear tort liability for the expanded damages. Users who maliciously induce the generation of infringing content shall bear corresponding liability.
• Injunctions for Personality Rights Infringement: To prevent AI-related torts and avert irreversible harm to rights and interests, rights holders may apply for court orders to halt relevant conduct.
• AI Product Liability: Strictly determined in accordance with the Product Quality Law, with emphasis on whether producers have provided adequate explanations and warnings regarding applicable scenarios, inherent limitations, and risks.
• Regulating “Big Data Price Discrimination” and “AI-Impersonated Celebrity Endorsements”: Operators who use algorithms to impose unreasonable differential treatment shall be held liable for tort. Where AI is used to impersonate celebrities in product promotions and constitutes fraud, consumers’ claims for punitive damages shall be supported.
• Liability for Autonomous Vehicle Accidents: Clear distinctions are drawn between product defects and driver fault, with refined rules for apportioning liability to support consumers’ lawful rights protection.

3. Adjudication of AI-Related Intellectual Property Disputes (Articles 12 – 16)

To balance innovation incentives with rights protection, the Opinions set forth specific provisions in five key areas:

• Infringement by AI-Generated Content: Liability is determined by comprehensively considering factors such as the type of service, training data sources, and the degree of participation by each party. AI developers asserting a non-infringement defense must provide supporting evidence, including training data sources, process records, and scientific theoretical bases.
• Open-Source Software Liability: Appropriate liability exemptions are granted. Developers who provide code modules through free open-source channels and publicly disclose functionality and security risks may be exempted from tort liability.
• Patent Granting for AI-Related Inventions: Inventions that employ technical means to solve technical problems may be recognized as patentable subject matter. A natural person who uses AI to complete an invention and makes a substantive creative contribution may be recognized as the inventor.
• Performance of AI-Related Technology Contracts: Breach of contract liability is determined based on contractual terms, taking into account the characteristics of AI research and development.
• Data Use Standards: Lawfully acquired data rights and interests are protected by law. Compiled works are protected under copyright law, and trade secrets are protected under the Anti-Unfair Competition Law. Acts that undermine AI operational security—such as fabricating interfering data or conducting adversarial example attacks—shall incur legal liability.

4. Procedural Rules and Criminal Sanctions (Articles 17 – 20)

• Fact-Finding and Evidence Review: Courts shall strengthen litigation guidance and leverage the role of technical investigators in assisting fact-finding. Differentiated review priorities are established for electronic data, big data reports, and blockchain-based evidence preservation. When AI-generated content is submitted as evidence, courts shall comprehensively consider factors such as prompt design, degree of similarity, and consistency across repeated tests.
• Regulating Improper AI-Based Evidence Collection: The use of AI to fabricate evidence or falsify facts is strictly prohibited. Acts such as deleting or tampering with generation identifiers or selectively presenting results to obtain false evidence shall result in dismissal of claims and may incur fines or detention.
• Criminal Sanctions for AI-Related Crimes: Those who use AI to commit fraud, defamation, infringement of citizens’ personal information, or endanger traffic safety, where such acts constitute crimes, shall be held criminally liable in accordance with the law.

5. Trial Guidance Mechanisms (Articles 21 – 24)

The Opinions propose source-level governance through diversified dispute resolution mechanisms, strengthen the functions of higher courts in elevated jurisdiction, case guidance, and trial supervision to unify adjudication standards for similar cases. They also promote collaborative coordination among courts and departments such as cyberspace administration, public security, and market regulation, while deepening international judicial exchange and cooperation to build a full-chain governance system.

Practical Significance: Providing a Precise Benchmark for the AI Industry and Judicial Practice

The release of the Opinions is not merely an improvement of judicial adjudication rules; it represents a profound empowerment of both the compliant development of the AI industry and the practical operation of judicial work.

Compliance Guidance for AI Enterprises: Drawing Red Lines and Providing Safe Harbors

The Opinions offer AI enterprises a clear compliance roadmap. On one hand, they delineate legal “red lines” that must not be crossed—behaviors such as AI face-swapping/voice-cloning, doxxing, and big data price discrimination will face clear civil and even criminal liabilities. In the processing of data for model training, enterprises must strictly observe the boundaries of “reasonable scope” and respect individuals’ “right to refuse.”

On the other hand, the Opinions also provide enterprises with “safe harbors.” For example, developers of free open-source software who fully disclose risks are granted liability exemptions, encouraging the healthy development of the open-source ecosystem. In the area of product liability, the Opinions emphasize that as long as producers have provided adequate explanations and warnings regarding applicable scenarios, limitations, and risks, liability can be reasonably defined, thereby avoiding the unlimited expansion of product liability.

Guidance Value for Judicial Practice: Unifying Adjudication Standards and Resolving Evidentiary Challenges

The Opinions systematically construct, for the first time, an adjudication rule framework for AI-related disputes, resolving the long-standing problems of “lack of applicable law” or “difficulty in application” in judicial practice. By clarifying the fault liability principle and multi-dimensional criteria for determining fault, the Opinions provide an operable judicial benchmark for complex technological disputes.

Particularly in the area of evidence review, the Opinions explore and establish rules for examining AI-generated content as evidence. Addressing the pain point of “key evidence being in the hands of the defendant,” the Opinions design a “dual-track burden of proof” structure for developers’ non-infringement defenses, effectively resolving the evidentiary困境 faced by rights holders and laying a solid foundation for consistent judgments in similar cases.

In summary, the Opinions represent a landmark achievement in China’s judicial response to the AI era. By carefully balancing the protection of rights with the encouragement of innovation, they not only provide courts with clear and actionable adjudication standards but also offer the AI industry a predictable legal environment in which to thrive. As AI technology continues to evolve at an unprecedented pace, the Opinions serve as both a compass and a guardrail—ensuring that the journey of artificial intelligence remains firmly anchored in the principles of people-centeredness, innovation, and security.

¹ https://ipc.court.gov.cn/zh-cn/news/view-6033.html