On September 7, 2026, the Supreme People’ s Court formally issued the Opinions on Lawfully Adjudicating Cases Involving Artificial Intelligence Disputes (Fa Fa [2026] No. 10; hereinafter, the “Opinions” ).

In recent years, China’ s artificial intelligence (AI) industry has developed rapidly and become increasingly important. Meanwhile, AI-related disputes have continued to enter the judicial arena, with new types of cases constantly emerging, while China still lacks a comprehensive AI statute. Against this backdrop, the Supreme People’ s Court issued the Opinions. As the first document on adjudication rules for AI-related cases issued by China’ s highest judicial organ, the Opinions, based on existing law, systematically address AI-related torts, intellectual property, litigation procedures, criminal accountability, and other issues. In particular, the Opinions provide specific guidance for AI-related intellectual property disputes, offering clear compliance guidance and a basis for rights enforcement to technology and innovation enterprises, AI R&D entities, and intellectual property professionals.

I. Overview of the Opinions

The Opinions focus squarely on issues of broad public concern and set out targeted substantive adjudication and procedural rules for AI-related disputes. The Opinions comprise five parts and 24 articles, whose key content is briefly introduced below by part.

· Part I (Articles 1-2): Three Basic Principles

This part sets out three basic principles: people-centered adjudication that prevents algorithmic discrimination and protects individuals’ rights to know and choose in automated decision-making; support for innovation through an inclusive and prudent balance between rights protection and technological development, equally protecting the innovation interests of small and medium-sized enterprises; and a fortified security baseline that reasonably allocates liabilities among different actors.

· Part II (Articles 3-11): Determination of Tort Liability

This part first establishes the principle for attributing tort liability in AI-related cases: where the law does not expressly provide for no-fault liability or a presumption of fault liability, the fault liability principle shall apply. It also sets out rules for determining liability in scenarios involving personality rights, privacy rights, personal information rights and interests, consumer rights and interests, and other matters, and addresses the liability of generative AI service providers, AI product liability, compensation for traffic accidents involving autonomous driving, and other issues.

· Part III (Articles 12-16): Intellectual Property Disputes

This part specifically sets out adjudication rules for AI-related intellectual property disputes, covering five areas: liability for infringement involving AI-generated content, liability for open-source software, patent grant and validity determination, performance of technology contracts, and data compliance. It seeks to strike a balance between encouraging AI technological innovation and safeguarding the lawful rights and interests of rights holders.

· Part IV (Articles 17-20): Rules of Adjudicatory Procedure

This part improves the rules for fact-finding and evidentiary review in AI-related cases, specifying differentiated focuses for reviewing new forms of evidence such as electronic data, big-data analysis reports, and blockchain-based evidence preservation. It also provides for punishment for conduct that obstructs judicial order, such as using AI to fabricate evidence or bring sham litigation; where such conduct constitutes a crime, criminal liability shall be pursued in accordance with law. 

· Part V (Articles 21-24): Judicial Work Mechanisms

This part emphasizes the importance of diversified dispute-resolution mechanisms, adjudicative supervision and guidance by higher courts, multi-department coordination and cooperation, and international judicial exchanges and cooperation.

II. Key Interpretation: Adjudication Rules for AI-Related Intellectual Property Disputes

Intellectual property protection is a major focus of the Opinions. A dedicated chapter is established to regulate the adjudication of intellectual property disputes in the AI field, focusing on five core scenarios: infringement involving AI-generated content, open-source software liability, patent grant and validity determination, performance of technology contracts, and data compliance. These rules are highly instructive for technology and innovation enterprises, R&D institutions, intellectual property professionals, and others. The provisions are reviewed and interpreted below, article by article.

1. Infringement Involving AI-Generated Content: Precise Allocation of Liability Among Multiple Parties (Article 12)

This article addresses copyright infringement and unfair competition. It clarifies that courts should determine the liability of AI developers, providers, and users based on factors such as the type of AI service, training data sources, each party’ s degree of participation, compliance measures, and profits obtained. Developers raising a non-infringement defense must provide supporting materials, including training data sources, training records, and the model’ s mode of operation. Users who knew or should have known of a pre-existing work and used AI to generate substantially similar content without reasonable defense shall bear liability. Using AI for false advertising, fabricated transactions, or counterfeiting may also constitute unfair competition.

2. Open-Source AI Software: Differentiated Liability Exemptions (Article 13)
For open-source software widely used in AI, the Opinions establish an inclusive and prudent liability standard. Courts will consider factors such as open-source license agreements, restrictions on rights, compliance measures, and information disclosure, and grant appropriate liability exemptions to compliant open-source actors. Where a code module for AI R&D is provided free of charge on an open-source basis with its functions and security risks publicly disclosed, the developer or provider may be found not liable if another person’ s use of that module results in infringement, thereby protecting the vitality of the open-source AI ecosystem.

3. AI-Related Patents: Clarifying Standards for Patentability and Inventor Determination (Article 14)

The Opinions clarify the scope of patent protection for AI-related inventions and creations, resolving issues that have long been debated in the industry.

First, regarding determination of patentable subject matter, this article makes clear that an AI-related invention or creation that employs technical means conforming to the laws of nature, solves a technical problem, and achieves a technical effect conforming to the laws of nature should be deemed subject matter protected by the Patent Law, except where it violates laws or social morality, harms the public interest, or no natural person has made a substantive contribution. This standard is consistent with the three-element framework of “technical problem - technical means - technical effect” in the current Guidelines for Patent Examination and with Article 5 of the Patent Law concerning circumstances in which patent rights shall not be granted. At the same time, by treating“no substantive contribution by a natural person” as an exclusion, it further reinforces the central role of human beings.

Second, regarding determination of inventor status, this article provides that a natural person who makes a substantive and creative contribution to an invention or creation completed by AI shall be recognized as an inventor. This is fully consistent with the Guidelines for Patent Applications for AI-Related Inventions (Trial) issued by the CNIPA at the end of 2024. While clarifying the boundary for determining inventorship in AI-assisted innovation scenarios, this rule once again emphasizes the patent system’ s core requirement of creative contribution by natural persons.

Third, regarding the standard for sufficient disclosure in the specification, this article’ s provision on sufficient disclosure is consistent with Article 26(4) of the Patent Law, namely that the standard is whether a person skilled in the relevant technical field is able to carry it out. In this regard, the latest revised Guidelines for Patent Examination, which took effect on January 1, 2026, also specify drafting requirements for the specification in scenarios involving AI model construction, training, application, and other circumstances, further detailing the standard for sufficient disclosure.

4. AI Technology Contracts: Prudent Determination of Liability for Breach of Contract (Article 15)

In view of the high uncertainty and rapid iteration characteristic of AI technology R&D, this article clarifies the adjudication rules for technology contract disputes: in determining liability for breach of contract, full consideration should be given to the characteristics of AI technology R&D, whether the technology developer has made reasonable efforts, and other relevant circumstances.

5. AI Data Compliance: Comprehensive Protection of Data-Related Intellectual Property Rights and Competitive Interests (Article 16)

For AI data lawfully acquired, processed, or derived, this article establishes a multi-layered data protection framework: data that meets the requirements for a compilation work is protected under the Copyright Law; data constituting a trade secret is protected under the Anti-Unfair Competition Law. At the same time, the article strictly regulates non-compliant conduct in the data field and specifies the legal liabilities for various forms of data-related noncompliance.

III. Practical Recommendations for Innovators

In light of the adjudication rules established by the Opinions and the judicial orientation reflected in AI-related intellectual property cases, we offer the following practical recommendations, from the perspectives of intellectual property strategy, R&D compliance, and contract risk control.

1. Intellectual Property Strategy: Regularize AI Patent Applications and Clarify Inventorship and Disclosure

Article 14 of the Opinions addresses three core issues in AI-related patent applications: patentable subject matter, inventorship, and sufficient disclosure. For patent practitioners, this provision is particularly significant because it consolidates the key thresholds that AI-related applications must meet.

First, enterprises are advised to frame AI-related inventions around technical problems, technical means, and technical effects, and avoid claiming pure algorithms, abstract rules, or purely intellectual activities. Where an invention involves an AI model, the application should focus on how the model is constructed, trained, and applied to solve a specific technical problem in a particular field.         

Second, inventorship should be attributed to natural persons who have made substantive and creative contributions to the invention. When enterprises use AI in technological R&D, they are advised to fully preserve evidence of creative contributions made by natural persons, including process documents concerning conception of technical solutions, parameter design, algorithm optimization, and solution selection, so as to avoid rejection of patent applications or ownership disputes due to the absence of substantive human contribution.

Third, in line with the disclosure requirements of the Opinions and Patent Examination Guidelines, patent specifications should be drafted to ensure that technical solutions are sufficiently disclosed and logically clear, with particular attention to the construction, training, and application of AI models. Overly general descriptions of model architecture, training data, and training processes should be avoided, and sufficient technical details should be provided to support the claimed scope.

2. R&D Compliance Management: Regularize Training Data and Open-Source Use to Mitigate Infringement Risks

Article 12 requires AI developers raising a non-infringement defense to provide supporting materials such as the sources of training data, records of the training process, and the mode's mode of operation. Enterprises should therefore prioritize lawfully authorized or undisputed public data, strictly complete authorization for personal information or third-party works, and retain compliance records for possible evidentiary needs. AI-generated content should undergo infringement screening and similarity checks before release, with filtering and review records retained.

Article 13 establishes conditional liability exemptions for open-source developers and providers. Accordingly, enterprises using open-source AI modules should review license terms, comply with disclosure requirements, maintain an open-source component inventory, and record code sources, scope of use, and modifications to support a liability exemption claim in disputes. 

3. Contract and IP Ownership Transaction Risk Control: Refine AI Technology Contract Terms and Clarify Liability for Breach

Under Article 15, AI technology contract disputes should consider AI R&D characteristics and whether the developer made reasonable efforts. Enterprises should therefore optimize contract terms for AI development, transfer, and licensing—specifying delivery standards, acceptance criteria, breach liability boundaries, and performance deviations caused by technological uncertainty. Enterprises should also clarify ownership, use rights, and revenue allocation in AI R&D and cooperation through contracts, and adopt layered protection combining patents, copyright, and trade secrets for algorithmic models, core code, training datasets, and application solutions.

IV. Conclusion

The issuance of the Opinions marks the systematic establishment of judicial adjudication rules in China’ s AI field. For innovation entities, it is important both to proactively align with the standards of the Opinions by maintaining compliance records and managing risks in advance, and to make effective use of rules such as open-source liability exemptions and the fault-liability principle to protect their rights and interests in accordance with law. We will continue to monitor the latest developments in judicial practice concerning AI-related intellectual property and provide you with timely and professional support.