August 2026
Compliance Tips for the Application of Artificial Intelligence Technology (IV) — Compliance in AI‑Generated Content (Mainland China)
In the previous articles of this series, we analyzed algorithm compliance and data processing compliance; in this final article, we will discuss issues relating to compliance in AI‑generated content.
The term “AI-generated content” as used in this article refers to the conduct described in the Interim Measures for Generative AI Services, namely “using generative artificial intelligence technology to provide the public within the territory with generated text, images, audio, video, and other content, which is subject to regulation under those Measures.” ［1］ From the perspective of judicial practice, AI-generated content may involve the following legal risks: (1) whether the generated content is lawful and labeled as required by law; (2) whether the generated content constitutes a copyrighted work that can be protected under copyright law; (3) whether the generated content infringes the copyright entitlement in another person’s copyrighted work; (4) whether the AI-generated content infringes other civil rights and interests of others; and (5) whether the use of other persons’ copyrighted works in the model training process constitutes infringement.
With respect to the legality of labeling AI-generated content, the author has previously discussed this issue in a dedicated article (see the author’s article “A Brief Introduction to the Measures for the Labeling of AI-Generated and Synthetic Content” ［2］ ), and will not repeat that discussion here. This article will primarily provide a brief discussion on the remaining issues.
I. Does AI‑generated content constitute a copyrighted work?
Under the laws of China, as a general rule, for something to be protected under copyright law, it must first satisfy the requirements for a copyrighted work as set forth in Article 3 of the Copyright Law, namely the following four elements: (1) whether it falls within the literary, artistic, or scientific domains; (2) whether it possesses originality; (3) whether it has a certain form of expression; and (4) whether it constitutes an intellectual achievement.
With respect to AI-generated content, according to existing judicial cases, the primary controversy lies in whether such AI-generated content meets the requirement of “originality.” In light of the adjudicatory standards reflected in judgments such as (2023) Jing 0491 Min Chu No. 11279, (2025) Gan 0602 Zhi Min Chu No. 47, (2025) Yue 1403 Min Chu No. 3215, and (2025) Yue 1403 Min Chu No. 3214, where the author, in the course of using artificial intelligence technology to generate content, demonstrates personalized expression, intellectual selection and judgment, aesthetic preferences, and an understanding of the artistic nature of the copyrighted work, and achieves a certain level of creative height, there is a relatively high likelihood that the content will be recognized as a copyrighted work. Conversely, where the AI-generated content highly depends on the technical functionalities of the AI software, the author has not invested substantial personal aesthetic judgment or creative effort, and the AI-generated content does not reflect the author’s unique intellectual creation and lacks originality and novelty, it will be difficult for such content to be recognized as a copyrighted work protected under the Copyright Law.
Therefore, for authors who wish to obtain copyright protection, attention should be paid to preserving evidence of intellectual input and creative traces in the course of creation, such as records of parameter adjustments, work logs, and comparisons of different generated versions, so as to avoid difficulties in asserting rights at a later stage. On the other hand, as reflected in the judgment (2023) Jing 0491 Min Chu No. 11279, in general, the end-user who actually operates the system is more likely to be recognized as the author rather than the service platform; accordingly, where a service platform plans any commercial use of end-users’ works, it should expressly provide for such use in advance in the user terms, so as to avoid disputes.
II. Does AI‑generated content infringe the copyright entitlement in other persons’ copyrighted works?
Under the laws of China, the copyright entitlement enjoyed by a copyright holder refers to the rights enjoyed under Article 10 of the Copyright Law according to the type of copyrighted work. Furthermore, pursuant to Articles 52 and 53 of the same law, where another person’s copyrighted work is used without the permission of the rights holder, such use may constitute infringement. If AI‑generated content involves plagiarism, alteration, reproduction, adaptation, or dissemination of another person’s copyrighted work online without obtaining the consent or license of the rights holder, it may constitute infringement under the Copyright Law.
For example, in the “Ultraman Images” case (2024) Zhe 0192 Min Chu No. 1587, the court held that “the alleged infringing content is highly similar to the rights holder’s copyrighted work in respect of the Ultraman character image, the color scheme, and the details of the costume, and as a whole constitutes substantial similarity,” and Ultraman is a well-known intellectual property, so the likelihood that the infringer had previously accessed the copyrighted work was extremely high; therefore, where AI-generated content is similar to other prior copyrighted works, if such AI-generated content is determined to meet the criteria of “access plus substantial similarity,” there is a very high likelihood that it will be found to constitute infringement.
In situations where AI-generated content created by end-users constitutes infringement, the service platform may also be found to have committed contributory infringement. According to the adjudicatory standards in the aforementioned “Ultraman Images” case, whether a service platform is deemed to have committed contributory infringement will mainly be assessed based on factors such as: the profit model of the service provider, the fame and influence of the rights holder’s copyrighted work, the obviousness of the infringing facts, the level of development of artificial intelligence technology, the feasibility and cost of alternative designs to avoid harm, the necessary measures that can be taken and their effectiveness, the impact of liability for infringement on the industry, and whether the service provider’s duty of care matches its information management capability, among others ［3］ .
Therefore, with respect to whether end‑users’ AI‑generated content may infringe the intellectual property rights of others, service platform providers should not only pay attention to including advance provisions on intellectual property infringement in the user terms, but should also consider whether technical means can be employed to proactively prevent similar infringements, thereby avoiding being held liable.
III. Does AI‑generated content infringe other civil rights and interests of others?
In addition to infringing other persons’ copyright entitlement, where AI‑generated content involves the use of another person’s portrait, voice, name, and the like without permission, it may constitute an infringement of that person’s portrait right, personality right, reputation right, and other such rights.
For example, in case (2021) Jing 04 Min Zhong No. 777, a bookkeeping software company used the portrait of a well-known individual, Mr. He, as one of the AI companions and encouraged end-users to interact with this AI persona. The court ultimately found that the company, by providing end-users with an AI virtual character in the software that contained Mr. He’s name, was engaging in commercial use of Mr. He’s name and thereby infringed his name right. Although the portrait image of Mr. He had been uploaded by an end-user, the company, without authorization from Mr. He, pushed within the software system the portrait image at issue that had been uploaded and created by the end-user, and its mechanism for reviewing training corpora did not filter out infringing content; it even encouraged end-users to create corpora involving Mr. He’s portrait image, and in practice used and publicly displayed Mr. He’s portrait, thereby infringing his portrait right.
Similarly, in case (2025) Chuan 7101 Min Chu No. 8546, the plaintiff, Mr. Sun, is a male film and television actor from mainland China who has appeared in multiple film and television works and enjoys a certain degree of fame. On February 9, 2024, the defendant published a short dynamic video using the plaintiff’s portrait and an AI-generated voice. The court ultimately held that, with respect to voice processed by artificial intelligence technology, so long as the general public or the public within a certain scope can, based on timbre, tone, and manner of articulation, identify a specific natural person, that natural person’s voice-related rights extend to such AI-generated voice, and the defendant’s use of his voice without permission constituted an infringement of his voice right.
Moreover, using AI software to generate images that vilify or mock others will also constitute an infringement of others’ name right and portrait right. ［4］
IV. Does the use of other persons’ copyrighted works in the model training process constitute infringement?
Pursuant to Article 7 of the Interim Measures for Generative AI Services, providers of generative AI services shall lawfully conduct training data processing activities such as pre‑training and optimization training, and, where such activities involve intellectual property rights, shall not infringe intellectual property rights lawfully enjoyed by others. However, in practice, as to how to determine whether the intellectual property rights of others have been infringed, judicial practice appears to have not yet formed a uniform conclusion.
From the perspective of the provisions on fair utilization under the Copyright Law, where a party intends to rely on fair utilization under Article 24 of the Copyright Law as a defense, at a minimum the following elements must be satisfied: the author’s identity must be indicated, the normal use of the copyrighted work must not be affected, and the lawful rights and interests of the copyright holder must not be unreasonably prejudiced. In light of the actual operation of the model training process, whether these elements can be satisfied remains open to discussion.
On the other hand, in the aforementioned “Ultraman Images” case, the trial court also expressed certain views on the use of copyright in the model training process: “Generative artificial intelligence mainly includes four important stages: data input, data training, content output, and content usage. From the perspective of promoting the development of generative artificial intelligence technology, with respect to determining whether the data input and data training activities of large models constitute infringement, it is appropriate to adopt a relatively lenient and tolerant standard of determination, while with respect to determining whether the content output and content usage activities of large models constitute infringement, it is appropriate to adopt a relatively stricter standard of determination, so as to achieve a balance between development and protection through differentiated measures.” However, this remains an adjudicatory standard in an individual case, and it cannot be ruled out that more lenient adjudicatory standards may be applied in similar cases in the future.
In summary, this article has briefly analyzed the potential legal risks associated with AI‑generated content under current circumstances. In light of the adjudicatory standards already in place, in situations where AI‑generated content constitutes a work, is substantially similar to prior copyrighted works, or involves the use of another person’s name, portrait, or voice, attention should be paid to reviewing one’s own compliance obligations so as to avoid being held liable.
［1］ See Article 2 of the Interim Measures for Generative AI Services: “These Measures apply to services (hereinafter referred to as ‘generative AI services’) that use generative artificial intelligence technology to provide the public within the territory of the People’s Republic of China with generated text, images, audio, video, and other content.”
Where the state has separate provisions for engaging in activities such as news and publishing, film and television production, and literary and artistic creation by using generative AI services, such provisions shall prevail.
Where industry organizations, enterprises, educational and research institutions, public cultural institutions, and relevant professional institutions engage in the research and development and application of generative artificial intelligence technology but do not provide generative AI services to the public within the territory, the provisions of these Measures do not apply.
［2］ Introduction to the Measures for Labeling AI-Generated or Composed Content (Mainland China)
［3］ See Case No. 29 “Determination of Infringement Liability of Generative AI Service Providers” in the Summary of the 2024 Annual Report on Legal Application Issues in Intellectual Property Cases by Courts at the National Level.
［4］ See the case of Cheng v. Sun concerning liability for online infringement — Typical Case No. 6 involving artificial intelligence published by the Beijing Internet Court.
The term “AI-generated content” as used in this article refers to the conduct described in the Interim Measures for Generative AI Services, namely “using generative artificial intelligence technology to provide the public within the territory with generated text, images, audio, video, and other content, which is subject to regulation under those Measures.” ［1］ From the perspective of judicial practice, AI-generated content may involve the following legal risks: (1) whether the generated content is lawful and labeled as required by law; (2) whether the generated content constitutes a copyrighted work that can be protected under copyright law; (3) whether the generated content infringes the copyright entitlement in another person’s copyrighted work; (4) whether the AI-generated content infringes other civil rights and interests of others; and (5) whether the use of other persons’ copyrighted works in the model training process constitutes infringement.
With respect to the legality of labeling AI-generated content, the author has previously discussed this issue in a dedicated article (see the author’s article “A Brief Introduction to the Measures for the Labeling of AI-Generated and Synthetic Content” ［2］ ), and will not repeat that discussion here. This article will primarily provide a brief discussion on the remaining issues.
I. Does AI‑generated content constitute a copyrighted work?
Under the laws of China, as a general rule, for something to be protected under copyright law, it must first satisfy the requirements for a copyrighted work as set forth in Article 3 of the Copyright Law, namely the following four elements: (1) whether it falls within the literary, artistic, or scientific domains; (2) whether it possesses originality; (3) whether it has a certain form of expression; and (4) whether it constitutes an intellectual achievement.
With respect to AI-generated content, according to existing judicial cases, the primary controversy lies in whether such AI-generated content meets the requirement of “originality.” In light of the adjudicatory standards reflected in judgments such as (2023) Jing 0491 Min Chu No. 11279, (2025) Gan 0602 Zhi Min Chu No. 47, (2025) Yue 1403 Min Chu No. 3215, and (2025) Yue 1403 Min Chu No. 3214, where the author, in the course of using artificial intelligence technology to generate content, demonstrates personalized expression, intellectual selection and judgment, aesthetic preferences, and an understanding of the artistic nature of the copyrighted work, and achieves a certain level of creative height, there is a relatively high likelihood that the content will be recognized as a copyrighted work. Conversely, where the AI-generated content highly depends on the technical functionalities of the AI software, the author has not invested substantial personal aesthetic judgment or creative effort, and the AI-generated content does not reflect the author’s unique intellectual creation and lacks originality and novelty, it will be difficult for such content to be recognized as a copyrighted work protected under the Copyright Law.
Therefore, for authors who wish to obtain copyright protection, attention should be paid to preserving evidence of intellectual input and creative traces in the course of creation, such as records of parameter adjustments, work logs, and comparisons of different generated versions, so as to avoid difficulties in asserting rights at a later stage. On the other hand, as reflected in the judgment (2023) Jing 0491 Min Chu No. 11279, in general, the end-user who actually operates the system is more likely to be recognized as the author rather than the service platform; accordingly, where a service platform plans any commercial use of end-users’ works, it should expressly provide for such use in advance in the user terms, so as to avoid disputes.
II. Does AI‑generated content infringe the copyright entitlement in other persons’ copyrighted works?
Under the laws of China, the copyright entitlement enjoyed by a copyright holder refers to the rights enjoyed under Article 10 of the Copyright Law according to the type of copyrighted work. Furthermore, pursuant to Articles 52 and 53 of the same law, where another person’s copyrighted work is used without the permission of the rights holder, such use may constitute infringement. If AI‑generated content involves plagiarism, alteration, reproduction, adaptation, or dissemination of another person’s copyrighted work online without obtaining the consent or license of the rights holder, it may constitute infringement under the Copyright Law.
For example, in the “Ultraman Images” case (2024) Zhe 0192 Min Chu No. 1587, the court held that “the alleged infringing content is highly similar to the rights holder’s copyrighted work in respect of the Ultraman character image, the color scheme, and the details of the costume, and as a whole constitutes substantial similarity,” and Ultraman is a well-known intellectual property, so the likelihood that the infringer had previously accessed the copyrighted work was extremely high; therefore, where AI-generated content is similar to other prior copyrighted works, if such AI-generated content is determined to meet the criteria of “access plus substantial similarity,” there is a very high likelihood that it will be found to constitute infringement.
In situations where AI-generated content created by end-users constitutes infringement, the service platform may also be found to have committed contributory infringement. According to the adjudicatory standards in the aforementioned “Ultraman Images” case, whether a service platform is deemed to have committed contributory infringement will mainly be assessed based on factors such as: the profit model of the service provider, the fame and influence of the rights holder’s copyrighted work, the obviousness of the infringing facts, the level of development of artificial intelligence technology, the feasibility and cost of alternative designs to avoid harm, the necessary measures that can be taken and their effectiveness, the impact of liability for infringement on the industry, and whether the service provider’s duty of care matches its information management capability, among others ［3］ .
Therefore, with respect to whether end‑users’ AI‑generated content may infringe the intellectual property rights of others, service platform providers should not only pay attention to including advance provisions on intellectual property infringement in the user terms, but should also consider whether technical means can be employed to proactively prevent similar infringements, thereby avoiding being held liable.
III. Does AI‑generated content infringe other civil rights and interests of others?
In addition to infringing other persons’ copyright entitlement, where AI‑generated content involves the use of another person’s portrait, voice, name, and the like without permission, it may constitute an infringement of that person’s portrait right, personality right, reputation right, and other such rights.
For example, in case (2021) Jing 04 Min Zhong No. 777, a bookkeeping software company used the portrait of a well-known individual, Mr. He, as one of the AI companions and encouraged end-users to interact with this AI persona. The court ultimately found that the company, by providing end-users with an AI virtual character in the software that contained Mr. He’s name, was engaging in commercial use of Mr. He’s name and thereby infringed his name right. Although the portrait image of Mr. He had been uploaded by an end-user, the company, without authorization from Mr. He, pushed within the software system the portrait image at issue that had been uploaded and created by the end-user, and its mechanism for reviewing training corpora did not filter out infringing content; it even encouraged end-users to create corpora involving Mr. He’s portrait image, and in practice used and publicly displayed Mr. He’s portrait, thereby infringing his portrait right.
Similarly, in case (2025) Chuan 7101 Min Chu No. 8546, the plaintiff, Mr. Sun, is a male film and television actor from mainland China who has appeared in multiple film and television works and enjoys a certain degree of fame. On February 9, 2024, the defendant published a short dynamic video using the plaintiff’s portrait and an AI-generated voice. The court ultimately held that, with respect to voice processed by artificial intelligence technology, so long as the general public or the public within a certain scope can, based on timbre, tone, and manner of articulation, identify a specific natural person, that natural person’s voice-related rights extend to such AI-generated voice, and the defendant’s use of his voice without permission constituted an infringement of his voice right.
Moreover, using AI software to generate images that vilify or mock others will also constitute an infringement of others’ name right and portrait right. ［4］
IV. Does the use of other persons’ copyrighted works in the model training process constitute infringement?
Pursuant to Article 7 of the Interim Measures for Generative AI Services, providers of generative AI services shall lawfully conduct training data processing activities such as pre‑training and optimization training, and, where such activities involve intellectual property rights, shall not infringe intellectual property rights lawfully enjoyed by others. However, in practice, as to how to determine whether the intellectual property rights of others have been infringed, judicial practice appears to have not yet formed a uniform conclusion.
From the perspective of the provisions on fair utilization under the Copyright Law, where a party intends to rely on fair utilization under Article 24 of the Copyright Law as a defense, at a minimum the following elements must be satisfied: the author’s identity must be indicated, the normal use of the copyrighted work must not be affected, and the lawful rights and interests of the copyright holder must not be unreasonably prejudiced. In light of the actual operation of the model training process, whether these elements can be satisfied remains open to discussion.
On the other hand, in the aforementioned “Ultraman Images” case, the trial court also expressed certain views on the use of copyright in the model training process: “Generative artificial intelligence mainly includes four important stages: data input, data training, content output, and content usage. From the perspective of promoting the development of generative artificial intelligence technology, with respect to determining whether the data input and data training activities of large models constitute infringement, it is appropriate to adopt a relatively lenient and tolerant standard of determination, while with respect to determining whether the content output and content usage activities of large models constitute infringement, it is appropriate to adopt a relatively stricter standard of determination, so as to achieve a balance between development and protection through differentiated measures.” However, this remains an adjudicatory standard in an individual case, and it cannot be ruled out that more lenient adjudicatory standards may be applied in similar cases in the future.
In summary, this article has briefly analyzed the potential legal risks associated with AI‑generated content under current circumstances. In light of the adjudicatory standards already in place, in situations where AI‑generated content constitutes a work, is substantially similar to prior copyrighted works, or involves the use of another person’s name, portrait, or voice, attention should be paid to reviewing one’s own compliance obligations so as to avoid being held liable.
［1］ See Article 2 of the Interim Measures for Generative AI Services: “These Measures apply to services (hereinafter referred to as ‘generative AI services’) that use generative artificial intelligence technology to provide the public within the territory of the People’s Republic of China with generated text, images, audio, video, and other content.”
Where the state has separate provisions for engaging in activities such as news and publishing, film and television production, and literary and artistic creation by using generative AI services, such provisions shall prevail.
Where industry organizations, enterprises, educational and research institutions, public cultural institutions, and relevant professional institutions engage in the research and development and application of generative artificial intelligence technology but do not provide generative AI services to the public within the territory, the provisions of these Measures do not apply.
［2］ Introduction to the Measures for Labeling AI-Generated or Composed Content (Mainland China)
［3］ See Case No. 29 “Determination of Infringement Liability of Generative AI Service Providers” in the Summary of the 2024 Annual Report on Legal Application Issues in Intellectual Property Cases by Courts at the National Level.
［4］ See the case of Cheng v. Sun concerning liability for online infringement — Typical Case No. 6 involving artificial intelligence published by the Beijing Internet Court.
Related Articles
Compliance Tips for the Application of Artificial Intelligence Technology (I) – Overview of AI Compliance Framework (Mainland China) Compliance Tips for the Application of Artificial Intelligence Technology (II) — Algorithm Compliance (Mainland China) Compliance Tips for the Application of Artificial Intelligence Technology (III) — Data Processing Compliance (Mainland China)
The contents of all newsletters of Shanghai Lee, Tsai & Partners (Content) available on the webpage belong to and remain with Shanghai Lee, Tsai & Partners. All rights are reserved by Shanghai Lee, Tsai & Partners, and the Content may not be reproduced, downloaded, disseminated, published, or transferred in any form or by any means, except with the prior permission of Shanghai Lee, Tsai & Partners.
The Content is for informational purposes only and is not offered as legal or professional advice on any particular issue or case. The Content may not reflect the most current legal and regulatory developments. Shanghai Lee, Tsai & Partners and the editors do not guarantee the accuracy of the Content and expressly disclaim any and all liability to any person in respect of the consequences of anything done or permitted to be done or omitted to be done wholly or partly in reliance upon the whole or any part of the Content. The contributing authors' opinions do not represent the position of Shanghai Lee, Tsai & Partners. If the reader has any suggestions or questions, please do not hesitate to contact Shanghai Lee, Tsai & Partners.
