On September 7, 2026, the Supreme People's Court held a press conference on the Opinions of the Supreme People's Court on Adjudicating Disputes Involving Artificial Intelligence in Accordance with the Law. Tao Kaiyuan, Vice-President of the Supreme People's Court; Zhou Jiahai, Director of the Research Office; Li Jian, Presiding Judge of the No. 3 Civil Division; and Si Yanli, Deputy Director of the Research Office, attended the conference and answered questions from journalists. The conference was chaired by Ji Zhongbiao, spokesperson of the Supreme People's Court.

Q: We have noted that the Opinions of the Supreme People's Court on Adjudicating Disputes Involving Artificial Intelligence in Accordance with the Law (hereinafter the "Opinions") are grounded in the current hot and difficult issues and respond in a targeted manner to matters of concern to the public and to the artificial intelligence industry. May we ask, in terms of the specific arrangement of its content, how do the Opinions properly handle the relationship between the protection of rights and interests and innovative development?

A: In his keynote speech at the opening ceremony of the 2026 World Artificial Intelligence Conference and the High-Level Meeting on Global AI Governance, General Secretary Xi Jinping pointed out that we must always give equal emphasis to development and security, grasp the trends and laws of AI development, continuously improve the relevant laws and regulations, policies and systems, application norms, and ethical guidelines, and ensure that AI is safe, reliable, and controllable, so that this fine steed of AI runs both fast and steadily. The Opinions are developed closely around the principle of "giving equal emphasis to development and security," focusing both on regulating in accordance with the law the illegal chaos of the misuse and malicious use of AI and on strengthening judicial protection of people's livelihood rights and interests in the digital age, while at the same time clearly and firmly supporting scientific and technological innovation and industrial innovation and safeguarding the healthy development of AI. Specifically:

With regard to safeguarding people's livelihood rights and interests, the Opinions mainly provide as follows. First, they intensify judicial protection of personality rights and interests. At present, illegal acts such as AI face-swapping, AI voice imitation, and "online doxxing" occur from time to time, drawing great concern from all sectors of society. To this end, the Opinions set out clear norms for holding accountable acts that use AI to infringe the right to one's likeness, the right to reputation, the right to privacy, voice rights and interests, and the like, so as to give full play to the norm-setting and guiding function of judicial adjudication and to promote AI being oriented toward the good. Second, they protect personal information rights and interests in accordance with the law. Large-scale, high-quality training data is the cornerstone of the development of AI technology. In order to balance and give consideration to both the protection of personal information rights and interests and the promotion of the AI industry, the Opinions, in accordance with the Civil Code, the Personal Information Protection Law, and other laws, make clear that where, for the purpose of training AI models, personal information that has already been lawfully disclosed is processed within a reasonable scope and the individual has not expressly refused, this is generally not to be found to constitute infringement. However, where there is a significant impact on the individual's rights and interests, the consent of the individual shall be obtained in accordance with the law. Third, they regulate in accordance with the law acts that harm consumers' lawful rights and interests, such as "big data price discrimination against loyal customers" and "impersonating celebrities to sell goods." Recently, "big data price discrimination" triggered by AI algorithms has repeatedly come to light, and "impersonating celebrities to sell goods" has also attracted much attention. We cannot expect every consumer to have "piercing eyes"; the law must draw its sword in time to safeguard consumers' lawful rights and interests. To this end, the Opinions make clear that where, for the same goods or services, a business operator uses algorithms to apply unreasonable differential treatment in transaction terms such as transaction prices, infringing on the lawful rights and interests of others and causing damage, the people's court shall find in accordance with the law that it bears the corresponding tort liability; and where AI is used to "impersonate a celebrity to sell goods" and this constitutes fraud, and the consumer claims punitive damages, the people's court shall support this in accordance with the law.

With regard to supporting innovative development, the Opinions mainly provide as follows. First, they accurately grasp the principle of attributing liability for AI-related torts, so as to avoid discouraging enthusiasm for innovation by imposing overly heavy liability while AI technology and industry are still in the early stages of development. The Opinions make clear that, for legal liability arising from the use of AI to infringe civil rights and interests, the fault liability principle shall apply, except where the law expressly provides for the application of no-fault liability or presumption-of-fault liability. Second, they strictly determine what constitutes an "AI product" in accordance with the law, so as to avoid the over-expansion of product liability. The Opinions make clear that, strictly in accordance with the definition of "product" in the Product Quality Law, AI products shall be limited to products carried on a physical object, such as intelligent robots and self-driving vehicles, while excluding AI services that have no physical carrier. Third, they comprehensively and accurately implement the criminal policy of combining leniency with severity, and handle criminal cases involving AI in accordance with the law. On the one hand, the Opinions emphasize that, for criminal acts intentionally committed using AI that endanger national security, infringe citizens' rights and interests, or disrupt social order, there shall be resolute and severe punishment in accordance with the law; on the other hand, they require that, for innovative conduct in the research, development, and application of AI, matters be handled prudently in accordance with the law, adhering to the principles of legality of crime and punishment and of the correspondence between crime, responsibility, and punishment. In addition, as Vice-President Tao introduced earlier, the Opinions make special provisions for properly adjudicating AI-related intellectual property disputes in accordance with the law, and the relevant provisions also fully embody the philosophy of supporting innovative development along the track of the rule of law.

Q: New types of intellectual property disputes triggered by AI-generated content are occurring frequently and have drawn wide attention. May we ask, what new provisions do the Opinions contain regarding the handling of intellectual property disputes involving AI-generated content?

A: Articles 12 to 16 of the Opinions are provisions made specifically for AI-related intellectual property disputes, and among them Article 12 provides guidance on the issue of AI-generated content that you mentioned. The Outline of the 15th Five-Year Plan expressly proposes exploring the establishment of rules for the attribution of rights in AI-generated products and for determining the rights and responsibilities of developers, operators, and users. The recently issued Plan for Intellectual Property Protection and Utilization under the 15th Five-Year Plan by the State Council further emphasizes improving the copyright system to adapt to the development of AI. The Opinions implement the relevant deployments of the central authorities and respond in a timely manner to the concerns of society. Specifically, there are three aspects.

First, the Opinions clarify the factors to be considered in determining the liability of the relevant parties in cases where AI-generated content infringes copyright. Infringement by AI-generated content involves multiple parties, including AI technology developers, providers, and users. Where AI-generated content is sued for infringing another's copyright, the liability of the relevant parties cannot be excused on the ground that the allegedly infringing content was AI-generated; rather, the principle must be upheld that the assumption of liability matches the party's control capability and duty of care. As to the determination of control capability and duty of care, one should generally take into account, in a comprehensive manner, factors such as the technical and business-model characteristics of the AI involved in the case, the role played by the party in the content-generation process, which party "fed" the relevant training data, the necessary measures taken, and the profits obtained. On this basis, whether the party has subjective fault in the infringement and the degree of such fault are to be determined in accordance with the law, and the tort liability is to be reasonably apportioned.

Second, the Opinions clarify the rules of proof in disputes over copyright infringement by AI-generated content. Where a rights holder claims that an AI service developer or provider has infringed its copyright, it shall bear the burden of proving prima facie facts such as that the infringing content was generated by that AI and that it is substantially similar to the rights holder's work. Because AI technology is complex and opaque, the relevant evidence is not something others can obtain. Therefore, where the AI developer raises a defense of non-infringement, it shall bear the burden of proving facts such as the source of the model's training data, the training process, and the mode of operation, and where necessary shall also provide relevant scientific and theoretical grounds in support. Where an AI user knew or should have known of a prior work, uses AI to generate a work substantially similar to the prior work, and has no exemption such as fair use, infringement shall be found to be established.

Third, the Opinions regulate in accordance with the law acts of unfair competition committed using AI-generated content. Over the past two years, disputes involving the use of AI to engage in unfair competition have been increasing. Business operators use AI-generated text, images, videos, virtual character images, and the like to fabricate traffic and favorable reviews, and even carry out false advertising by using AI face-swapping to fabricate popular-science videos; such acts not only harm the lawful rights and interests of others but also seriously mislead consumers and disrupt the order of market competition, and shall be found in accordance with the law to constitute counterfeiting infringement, false advertising, and the like.

It should be noted that, during the drafting and deliberation of the Opinions, there were great differences of opinion on the copyrightability of AI-generated content and on the characterization of using others' works without permission to train large AI models, and understanding of these issues still needs to be further deepened. Therefore, the Opinions do not, for the time being, make provisions on these two issues. At present, AI technology is developing at breakneck speed, and new situations and new problems keep emerging. Going forward, we will continue to pay close attention to intellectual property protection issues in AI and other emerging fields, continuously recognize and grasp the underlying laws in practice, continuously summarize adjudication experience, improve adjudication rules, and strive to promote the formation of consensus.

Q: As AI iterates ever faster, technologies such as deepfakes, intelligent collection, and user profiling have greatly lowered the threshold for infringement, and the protection of personality rights and interests faces severe challenges. You mentioned just now that the Opinions make provisions on hot issues that infringe personality rights and interests, such as "AI face-swapping and voice imitation" and "online doxxing." Could you give a more specific introduction?

A: Personality rights bear on the human dignity of every person and are the most basic and most important rights of civil subjects. We emphasize that the protection of personality rights and interests and the development of the AI industry should be balanced with an inclusive and prudent attitude; however, inclusiveness does not mean conniving, and prudence does not mean laissez-faire. In order to respond to the concerns of society and guide adjudication practice, the Opinions make targeted provisions, distinguishing among different types, for acts that abuse and maliciously use AI technology to infringe personality rights and interests. These mainly include the following.

First, they regulate in accordance with the law acts such as "AI face-swapping and voice imitation" that use AI to infringe the right to one's likeness, the right to reputation, and voice rights and interests. In practice, the strongest reactions concern AI deepfakes. For example, AI face-swapping technology can indiscriminately "harvest" everyone's likeness, and ordinary people may also become "targets" of deepfakes; likewise, using AI voice-imitation technology, one can "steal a voice" for a few yuan, with an effect convincing enough to pass as genuine. Recently, the news media also exposed a black-market chain selling "AI one-click undressing" software and tutorials, which the public deeply detests. Article 4 of the Opinions clearly provides: where, without consent, AI is used to process a natural person's name, likeness, and the like to generate a virtual digital image identifiable as that natural person and this image is used or made public, this constitutes an infringement of personality rights and interests such as the right to one's name and the right to one's likeness; where, without consent, a natural person's voice is used as training material to imitate its timbre, intonation, and pronunciation style so as to generate an identifiable synthesized human voice, this constitutes an infringement of voice rights and interests; and where a virtual image or synthesized voice is manipulated to carry out improper conduct or to make untrue statements that lower another's social evaluation, this constitutes an infringement of the right to reputation, and the actor's tort liability shall be pursued in accordance with the law. In addition, in response to the socially prominent phenomenon of "using AI to resurrect the deceased," the Opinions also make special provisions to protect the personality interests of the deceased in accordance with the law.

Second, they regulate in accordance with the law acts that use AI to carry out "online doxxing" and "human-flesh searches" and thereby infringe others' right to privacy. "Online doxxing" and "human-flesh searches" are prominent manifestations of online violence; they not only seriously infringe the victim's right to privacy and seriously disturb the victim's peace of life, but also affect the public's sense of security and pollute the online ecosystem. At present, phenomena have emerged of using AI technology to track and analyze publicly available information such as a specific natural person's telephone number, online accounts, and social media, so as to obtain private information or to leak or disclose the private information thus obtained; for this, Article 5 of the Opinions makes specific provisions.

Third, they clarify the rules for applying the personality-rights infringement injunction system. Acts that use AI to infringe personality rights and interests have such characteristics as low cost, rapid dissemination, wide impact, and damage that is difficult to reverse; the traditional remedy model of bringing a lawsuit after the fact is not conducive to protecting the victim's lawful rights and interests in a timely and effective manner, so applying the personality-rights infringement injunction in accordance with the law is especially important. To this end, Article 8 of the Opinions makes this clear. For example, where a victim has evidence proving that he or she has been subjected to AI face-swapping to create "pornographic rumors," and failure to stop it in time would cause irreparable harm to his or her lawful rights and interests, he or she may apply to the people's court for a personality-rights infringement injunction. Upon review, where the conditions are met, the people's court may both order the network user in accordance with the law to stop the relevant infringing act and order the network service provider and the AI service provider in accordance with the law to take necessary measures in a timely manner, so as to prevent the further expansion of the harm.

It should be specially noted that, at present, phenomena of using technologies such as deepfakes to commit fraud, insult, defamation, infringement of citizens' personal information, and the production, sale, and dissemination of obscene materials occur from time to time. Where the relevant acts constitute crimes, they shall be resolutely and severely punished in accordance with the law and criminal liability shall be pursued; the Opinions make special provisions on this.

Q: At present, generative AI software is widely used, and the related civil disputes have also drawn social attention. I have noticed that, for the determination of infringement by generative AI, the Opinions introduce the "safe harbor rule." Could you give an in-depth introduction to this issue?

A: Article 1195 of the Civil Code provides for the "notice-and-takedown rule," commonly referred to as the "safe harbor rule." In simple terms, where a network user uses network services to commit an infringing act (for example, posting online to abuse another person), the rights holder, upon discovering it, has the right to notify the network service provider to take measures such as blocking or deletion; upon receiving a qualified notice, the network service provider shall take necessary measures in a timely manner, failing which it shall bear joint and several liability with the network user for the expanded portion of the damage. Conversely, if the network service provider, upon receiving the notice, takes necessary measures in a timely manner, it generally does not have to bear tort liability.

In practice, there is controversy over whether a generative AI service provider may claim the application, by reference, of the "safe harbor rule." Having studied the matter, we consider that although generative AI services differ from traditional network services, the two also have similarities in their operating mechanisms. For example, a generative AI service provider likewise organizes vast amounts of information and provides it to network users; the output of a large model depends on factors such as the earlier training data, the model parameters, and the prompts entered by the user; the AI service provider can hardly foresee in advance all the content entered by network users; and, given the enormous volume of generated content, it is impossible to require the AI service provider to predict, review, and intercept one by one whether the generated content is infringing. However, once the AI, due to its own hallucinations or the user's malicious inducement, generates infringing content, the generative AI service provider, upon notice by the rights holder, should and is in a position to take necessary measures; this shows that there is a legitimate and reasonable basis for applying the "safe harbor rule" by reference to generative AI service providers.

To this end, the Opinions make provisions on the determination of liability for "AI hallucinations" that infringe personality rights and for users' malicious inducement of AI to infringe personality rights, making clear that: where content automatically generated by generative AI infringes personality rights and interests and, upon notice by the rights holder, the AI service provider fails to take necessary measures in a timely manner, such as stopping the generation of the infringing content, it shall bear tort liability for the damage caused in accordance with the law; and where a network user, by means such as entering infringing prompts, maliciously induces generative AI to generate infringing content and causes damage to another, that network user shall bear tort liability in accordance with the law, and where, upon notice by the rights holder, the AI service provider fails to take necessary measures in a timely manner, such as stopping the generation of the infringing content and blocking the relevant generation instructions, and the rights holder requests that the network user and the AI service provider bear civil liability, the people's court shall support this in accordance with the law. The above provisions conform to the laws of technological development and to the legislative spirit of the Civil Code, embody the organic unity of supporting innovation and governance in accordance with the law, and were unanimously endorsed by all parties during the consultation process.

Q: The impact of AI is across all fields, and adjudication activities will also face new risks and challenges. For example, some litigation participants submit to the people's court AI-generated evidence and other litigation materials without verifying them before submission. How do the Opinions respond to such acts that use AI to obstruct judicial order?

A: Litigation participants shall follow the principle of good faith in litigation activities. With the popularization of generative AI technology, it is becoming increasingly common for litigation participants to use AI to assist in litigation. However, AI is merely an auxiliary tool, and litigation participants may not exempt themselves from their own legal liability on the grounds of "AI hallucinations" or technological neutrality. Litigation participants shall screen and verify the evidence and other litigation materials they submit to the court to ensure their authenticity and accuracy. In order to guide the public in the reasonable use of AI, prevent false materials from entering litigation proceedings, and effectively safeguard judicial order, the Opinions make provisions in three aspects.

First, they punish in accordance with the law the use of AI for false litigation. Where a party, exploiting the self-learning and autonomous-decision-making characteristics of AI, fabricates case facts to conduct false litigation through human intervention or misleading means such as deleting or tampering with the labeling of generated or synthesized content, entering specific instructions, selectively presenting results, or adversarial interference, the people's court shall dismiss its claims in accordance with the law and, depending on the severity of the circumstances, impose a fine or detention; where a crime is constituted, criminal liability shall be pursued in accordance with the law.

Second, they regulate in accordance with the law acts of using AI to fabricate evidence. The Opinions clearly provide that where a litigation participant or another person uses AI to fabricate evidence and obstruct the people's court's adjudication of a case, the matter shall be handled in accordance with Article 114 of the Civil Procedure Law, that is, depending on the severity of the circumstances, a fine or detention may be imposed, and where a crime is constituted, criminal liability shall be pursued in accordance with the law.

Third, they clearly provide that litigation participants bear the duty to verify and explain in respect of AI-generated litigation documents, case-search reports, and the like. At present, the people's courts have already discovered a number of false cases generated using AI, and the parties and their agents had not comprehensively verified them before submission to the court. In January of this year, the news media reported such incidents intensively, and the people's courts' case database also contains entered cases. On this basis, the Opinions clearly provide that where a litigation participant submits litigation documents, case-search reports, and other materials that were generated using AI, it shall, before submission to the court, carefully verify the authenticity and accuracy of the relevant laws, judicial interpretations, cases, and other content, truthfully explain at the time of submission the circumstances of using AI assistance, and bear liability in accordance with the law for the authenticity and accuracy of the relevant content. Judiciaries outside China also pay great attention to this issue; we have noted that some countries have specially issued guidelines emphasizing that agents bear the responsibility of ensuring the accuracy of the materials submitted to the court.

Source: Supreme People's Court of the People's Republic of China, 2026-09-08. Original: https://ipc.court.gov.cn/zh-cn/news/view-6032.html