China AI Law Brief

China Sets a National Benchmark for AI Disputes

Analysis of China’s AI Regulatory Developments

I. Executive Summary

On September 7, 2026, China’s Supreme People’s Court (SPC) issued the Opinions on the Lawful Adjudication of Disputes Involving AI (the “Opinions”), providing the first national judicial guidance for AI-related disputes. The twenty-four provisions direct courts to apply existing laws, including the Civil Code, Personal Information Protection Law, Copyright Law, Consumer Protection Law, and Civil Procedure Law, to new AI-related factual circumstances.

When AI causes harm, including deep-fake impersonation, defamatory statements generated by AI, privacy breaches, or copyright conflicts, who bears legal liability? Before the Opinions, Chinese courts confronting similar AI-related legal questions sometimes took different approaches and reached different outcomes. As courts face increasingly novel AI-related questions, the Opinions seek to promote greater consistency in applying existing laws. In this sense, the SPC is establishing a national judicial benchmark for AI disputes.

The Opinions also mark an important development in China’s AI governance. Earlier rules have focused primarily on how AI systems should be developed, provided, and used. The Opinions address what happens after disputes reach the courts. For companies, the practical consequences extend beyond litigation. AI governance in China increasingly requires both regulatory compliance and litigation readiness.

II. Background

The rapid development of China’s AI market provides important context for the new judicial guidance. China’s AI industry exceeded RMB 1.2 trillion (about US$176.7 billion) in 2025, an increase of about 40% from the previous year. By June 2026, China had more than 6,600 AI companies, representing approximately 15% of the global total, according to the China Academy of Information and Communications Technology. As AI becomes embedded in a wider range of commercial activities and interactions with individuals, AI-related disputes are also increasing.

In April 2026, the SPC disclosed that it was accelerating work on judicial guidance for AI-related disputes. Chinese courts concluded 908 cases involving data ownership and transactions in 2025, an increase of 25.6% from the previous year. Although these were not all AI cases, the figure illustrates the broader growth of disputes involving data and emerging technologies.

The Opinions respond to this evolving judicial landscape. They were issued as an SPC judicial policy document rather than legislation or formal judicial interpretation. They direct courts to apply existing laws to new AI-related factual circumstances.

The distinction from earlier AI regulation is important. China’s existing rules on algorithmic recommendation, deep synthesis, generative AI, AI ethics, AI agents, and anthropomorphic AI services focus primarily on managing risks as AI systems are developed and deployed. The Opinions address a different question: what happens when harm occurs, rights conflict, or a court must determine responsibility.

III. Analysis of the New Judicial Guidance

A. Fault Liability as the General Rule

One of the key features of the Opinions is the SPC’s approach to AI-related tort liability: ordinary fault liability becomes the default for AI-related torts. The SPC explained that this approach is intended to avoid imposing excessive liability while AI technology and industry remain at an early stage of development, which could discourage innovation.

China’s Civil Code recognizes ordinary fault liability, presumed-fault liability, and no-fault liability. Under ordinary fault liability, the plaintiff generally must establish the defendant’s fault and causation. Presumed fault and no-fault liability apply only where specifically provided by law.

The Opinions make ordinary fault liability under Article 1165(1) of China’s Civil Code the general rule for AI-related torts unless existing law provides otherwise. In determining fault, courts should consider AI-specific factors, including the system’s autonomy and technological transparency, potential risks, measures taken to prevent harm, and the parties’ ability to foresee and control those risks.

The U.S. has no comparable nationwide tort standard for AI. Courts generally apply existing state tort doctrines case by case. Recent litigation against Meta illustrates the role of fault-based liability. In March 2026, a California jury found Meta negligent in designing Instagram and concluded that its negligence was a substantial factor in harming a young user. China’s approach differs institutionally because the SPC has now given courts nationwide guidance that ordinary fault liability should be the default for AI-related torts.

B. Evidence and the Algorithmic Black Box

One of the important litigation aspects of the Opinions concerns access to evidence. AI disputes often involve technically complex information concentrated in the hands of the developer or provider. Article 17 provides certain mechanisms to address these difficulties. Where a party objectively cannot collect relevant evidence on its own, it may apply for the court to investigate and collect the evidence. The court may also consider applications for evidence preservation and use experts and technical investigators where necessary.

Where one party controls documentary evidence or electronic data and refuses without justified reason to produce it, the court may accept the opposing party’s assertion that the evidence would be unfavorable to the controlling party. This is not a new AI-specific discovery system, but its express application to AI disputes is important because algorithmic opacity can make claims difficult to prove.

The Opinions also impose more specific evidentiary responsibilities in AI copyright disputes. A rights holder alleging infringement must establish initial facts, such as that the challenged content was generated by the AI system and is substantially similar to the protected work. A developer asserting noninfringement, however, may be required to provide evidence concerning training-data sources, the training process, model operation, and, where necessary, relevant scientific theory.

C. Personality Rights and Generative AI

The Opinions devote substantial attention to AI-enabled infringements of personality rights, including face swapping, voice cloning, synthetic statements, recreation of deceased persons, and AI-assisted doxxing.

For generative AI providers, the Opinions apply a notice-and-response mechanism. If automatically generated content infringes personality rights and, after receiving a rights holder’s notice, the provider fails promptly to take necessary measures such as stopping generation of the infringing content, it may bear liability for the resulting harm. A user who maliciously uses infringing prompts to induce harmful output may also bear liability.

In its accompanying Q&A, the SPC explains that providers cannot realistically predict, review, and intercept every possible generated output in advance. The judicial focus therefore includes what happens after the provider receives notice and has the ability to take necessary measures.

D. Training Data and Unresolved Copyright Questions

For AI model training, processing personal information that individuals have made public, or that has otherwise been lawfully made public, generally will not constitute infringement if the processing remains within a reasonable scope and the individual has not expressly objected. In determining reasonable scope, courts should consider the purpose of processing, its relationship to the model’s functionality, the sensitivity of the information, potential effects on the individual, and reasonably anticipated uses. Consent remains required where processing significantly affects an individual’s rights and interests.

Copyright disputes receive particular attention, including evidentiary obligations concerning training-data sources and model operations. At the same time, the Opinions leave two important copyright questions unresolved: whether AI-generated content can itself qualify for copyright protection and whether unauthorized use of copyrighted works to train AI models constitutes copyright infringement. The SPC acknowledged significant disagreement over these issues during the drafting process and concluded that further judicial experience is needed.

E. Open-Source AI

The Opinions differentiate the liability of open-source developers and providers from that of downstream developers and providers in Article 13. Courts assessing their liability should consider factors including the type of open-source license, restrictions on rights, safety and compliance measures, and the extent of information disclosure. The Opinions call for appropriate liability exemptions for open-source developers and providers.

Where an open-source developer or provider makes certain code modules needed for AI software development available free of charge and publicly explains their functions and safety risks, a court may find that the developer or provider is not liable when another party’s use of those modules causes infringement.

The provision does not create blanket immunity. It allows courts to distinguish among participants based on their respective roles, conduct, safety measures, and disclosures. Downstream developers and providers therefore cannot assume that the open-source nature of a model or component eliminates their own potential responsibility.

F. Consumer Protection and Autonomous Vehicles

The Opinions apply existing consumer-protection principles to AI-mediated transactions, including unreasonable algorithmic differential treatment in pricing and other transaction terms. They also address AI-enabled celebrity impersonation in sales and permit consumer remedies where the conduct constitutes fraud.

For autonomous and assisted-driving vehicles, the Opinions direct courts to apply existing tort, road-traffic, and product-liability rules. Evidence such as driving event records may be necessary to determine causation and responsibility and may be controlled by manufacturers, sellers, or operators.

G. AI in Judicial Proceedings

The Opinions also address the use of generative AI in judicial proceedings. Parties using AI to prepare court filings or conduct case research must verify the authenticity and accuracy of cited laws, judicial interpretations, and cases, disclose their use of AI, and remain responsible for the materials they submit.

Using AI to fabricate evidence or create false litigation may result in fines, detention, or, where the conduct constitutes a crime, criminal liability. Courts are also given guidance for evaluating AI-generated evidence. The principle is straightforward: AI may assist people in judicial proceedings, but it does not replace human responsibility.

H. Cross-Border Questions Remain

The Opinions provide national adjudicatory guidance but do not resolve every cross-border issue involving offshore AI services, developers, datasets, or technical records. Questions involving jurisdiction, access to evidence stored abroad, cross-border transfers of litigation-related data, and enforcement may require analysis under other Chinese laws and procedural rules.

IV. Key Compliance Takeaways

Companies and organizations developing, providing, or deploying AI in China should consider six priorities:

• Review AI risk-management practices against the SPC’s fault-liability factors, including foreseeability, control, transparency, and preventive measures.
• Design recordkeeping with litigation in mind. Training-data provenance, model versions, testing records, technical logs, risk assessments, and incident-response materials may later become evidence.
• Establish procedures for receiving and responding to notices involving AI-generated infringements of personality rights.
• Review personal information and copyrighted materials used in model training, maintain relevant provenance and licensing records, and continue monitoring copyright developments that the Opinions deliberately leave unresolved.
• For open-source AI, document license terms, functionality, safety risks, disclosures, and downstream modifications rather than assuming that open-source status determines liability.
• Prepare for litigation and cross-border evidence issues by establishing appropriate AI-use verification procedures and understanding where relevant data and technical records are stored and how they may be accessed.

The broader significance of the Opinions is that China is beginning to build a common judicial framework for AI disputes nationwide. For companies, the Opinions matter well before a dispute reaches court. How a company manages training data, implements safeguards, maintains technical records, and responds to infringement complaints may later affect a court’s assessment of fault, evidence, and responsibility.

China’s approach also offers global privacy and AI governance professionals a useful point of comparison. The EU has built much of its AI framework through comprehensive legislation, while the U.S. relies on a combination of existing laws, state regulation, and case-by-case litigation. China is now adding nationwide judicial guidance on how existing laws should apply to AI disputes.

These differences matter for companies assessing cross-border AI liability. China’s AI governance is entering a new stage: regulators shape how AI is developed and used, while the SPC provides a national judicial benchmark for determining responsibility when AI causes harm.

About the Author

Sarah Zhao is a Partner at Rimon Law, where she advises multinational companies and other global businesses on AI, data protection, privacy, cybersecurity, cross-border data transfers, technology transactions, and regulatory compliance. Fluent in both English and Chinese, she combines a deep understanding of different legal systems and cultural subtleties with experience advising clients on complex cross-border legal and compliance issues.

China AI Law Brief is a periodic publication providing practical analysis of significant developments in China’s AI regulatory framework and their implications for companies, industry associations, government agencies, and international organizations.

(This publication is provided for informational purposes only and does not constitute legal advice. Readers should consult legal counsel regarding their specific circumstances.)

© 2026 Sarah Zhao, Partner, Rimon Law. All rights reserved.