On August 20, 2025, the Beijing Internet Court (“the Court”) published details of an infringement case regarding the use of an AI deepfake of a celebrity voice to sell goods by a seller on its official media account.
It is reported that the Plaintiff is a person with a certain degree of fame and social influence in the field of education and parenting. In 2024, the Plaintiff discovered that the Defendant, a culture media company, was promoting and selling multiple home education books on its operated online shops by using the Plaintiff’s public lectures and teaching videos with AI-generated voices that were highly similar to the Plaintiff’s. Therefore, the Plaintiff sued the Defendant in the Court.
The Plaintiff claimed that the Defendant used his image and the AI-generated voice without authorization to promote products while selling books on its online shop, which “closely associated the Plaintiff’s personal image with its commercially promoted subjects and caused the consumers to misidentify the Plaintiff as the agent or promoter of the sold books thereof. Such use of the Plaintiff’s personal image, professional background, and social influence to attract attention and increase business opportunity had infringed the Plaintiff’s rights of image and voice.” The Plaintiff also claimed that there was an entrustment relation between the Defendant as a seller and the video publisher (i.e., the live-streaming host), and they both jointly carried out the sales activities. The Defendant should have the obligation and capability to review the videos published by said host and therefore be liable to apologize, compensate the loss, and so on, for the publishing of the videos in question.
The Defendant replied that it was merely an operator of the involved online shop and a seller of the books in question; the promoted videos involved were published by other network users and not manufactured and published by itself; thus, it was not the subject of the infringement. The Defendant also argued that, as a public figure, the Plaintiff should have a certain degree of tolerance; the books promoted in the videos in question are good books that had neither damaged the Plaintiff's social image nor caused him economic loss. Therefore, it disagreed with the Plaintiff's claim.
Upon trial, the Court considered that the videos in question used the Plaintiff’s image, as well as the AI-generated voices, which are highly consistent with that of the Plaintiff in terms of tone, intonation, and pronunciation style; and therefore, it recognized that “an audience in a certain scope can establish a one-on-one corresponding relation between the involved AI-generated voice and the Plaintiff.” Considering the Plaintiff's reputation in the fields of education and parenting and the fact that the video was promoting books on family education, the Court found that the public viewing the video would easily associate the content of the video with the Plaintiff. Therefore, the voice in the video fell within the scope of protection of the Plaintiff's voice right. In addition, since the video also used the Plaintiff's image without authorization and synthesized and simulated the Plaintiff's voice, such acts constituted an infringement of the Plaintiff's image right and voice right.
As to the liability of the Defendant, the Court considered that although there was an entrustment relation between the Defendant and the promoter (i.e., the live-streaming host), they both published the videos in question and obtained relevant profit for the purpose of promoting the Defendant’s books in accordance with the platform rules and relevant service agreement, and the Defendant possessed the ability to review and manage the said videos on the basis of the platform rules and management access. The Court also considered that under the circumstance of massive use of the Plaintiff’s image and synthesized and simulated voice, the Defendant should have predicted that the videos had the risk of infringement and conducted a reasonable examination of whether the said videos obtained the authorization or not; however, due to failure to perform the reviewing obligations as it should be, the Defendant should be jointly liable for the infringed videos with the promoter. The Plaintiff has the right to choose the Defendant to be liable for all the compensation, which was supported by the Court. Accordingly, the Court ordered the Defendant to “apologize to the Plaintiff, compensate the economic loss and the reasonable cost for maintaining the right to RMB 120,000, and rejected the Plaintiff’s other requests.”
It is reported that the first instance judgment has come into effect and the Defendant has performed its obligation under the said judgment.
This is a typical case regarding the infringement of personal rights; however, it is also quite special, as it does not only involve the protection of the right of image but also the legal protection of the right of voice in the context of AI synthesis technology. In fact, the relevant content can be found in its legal basis in the PRC Civil Code. In accordance with Article 1019.1 of the Civil Code, “No organization or individual may infringe upon any other person's right to image by vilifying, defacing, or forging by means of information technology or otherwise. Without the consent of a person entitled to an image, the image of such person shall not be produced, used, or made public, unless otherwise provided by law.” Further, Article 1023.2 of the Civil Code provides that “The relevant provisions on protection of right to image shall apply as reference to protection of voice of natural person.”
In this case, the court further specified through the actual case that “natural human voices synthesized using AI deep synthesis technology are identifiable if the general public or relevant professionals can recognize a specific natural person based on their tone, intonation, pronunciation style, etc., and should therefore be included in the scope of protection of that natural person's voice right.” This ruling not only helps deepen our understanding of the application of the above provisions of the Civil Code but also provides important reference rules for similar cases in the future.
It is also worth noting that as the AI-generated synthetic images and sounds become increasingly realistic, this court ruling also imposes higher due diligence obligations on businesses and platforms that utilize such technologies for promotional purposes. For businesses, this means they should review the sources of promotional materials within reasonable limits (e.g., whether they have obtained authorization from rights holders) and retain evidence of fulfilling their review obligations. For platforms, this means they should label the AI-generated content in accordance with relevant laws and regulations and establish complaint mechanisms to better address new types of infringement, such as the unauthorized use of AI to infringe on image right or voice right.
In a word, this case not only provides important reference value for the effective protection of image right and voice right in the context of AI-generated synthetic content but also offers useful guidance on how to regulate the development of AI-generated synthetic technology.
It is reported that the Plaintiff is a person with a certain degree of fame and social influence in the field of education and parenting. In 2024, the Plaintiff discovered that the Defendant, a culture media company, was promoting and selling multiple home education books on its operated online shops by using the Plaintiff’s public lectures and teaching videos with AI-generated voices that were highly similar to the Plaintiff’s. Therefore, the Plaintiff sued the Defendant in the Court.
The Plaintiff claimed that the Defendant used his image and the AI-generated voice without authorization to promote products while selling books on its online shop, which “closely associated the Plaintiff’s personal image with its commercially promoted subjects and caused the consumers to misidentify the Plaintiff as the agent or promoter of the sold books thereof. Such use of the Plaintiff’s personal image, professional background, and social influence to attract attention and increase business opportunity had infringed the Plaintiff’s rights of image and voice.” The Plaintiff also claimed that there was an entrustment relation between the Defendant as a seller and the video publisher (i.e., the live-streaming host), and they both jointly carried out the sales activities. The Defendant should have the obligation and capability to review the videos published by said host and therefore be liable to apologize, compensate the loss, and so on, for the publishing of the videos in question.
The Defendant replied that it was merely an operator of the involved online shop and a seller of the books in question; the promoted videos involved were published by other network users and not manufactured and published by itself; thus, it was not the subject of the infringement. The Defendant also argued that, as a public figure, the Plaintiff should have a certain degree of tolerance; the books promoted in the videos in question are good books that had neither damaged the Plaintiff's social image nor caused him economic loss. Therefore, it disagreed with the Plaintiff's claim.
Upon trial, the Court considered that the videos in question used the Plaintiff’s image, as well as the AI-generated voices, which are highly consistent with that of the Plaintiff in terms of tone, intonation, and pronunciation style; and therefore, it recognized that “an audience in a certain scope can establish a one-on-one corresponding relation between the involved AI-generated voice and the Plaintiff.” Considering the Plaintiff's reputation in the fields of education and parenting and the fact that the video was promoting books on family education, the Court found that the public viewing the video would easily associate the content of the video with the Plaintiff. Therefore, the voice in the video fell within the scope of protection of the Plaintiff's voice right. In addition, since the video also used the Plaintiff's image without authorization and synthesized and simulated the Plaintiff's voice, such acts constituted an infringement of the Plaintiff's image right and voice right.
As to the liability of the Defendant, the Court considered that although there was an entrustment relation between the Defendant and the promoter (i.e., the live-streaming host), they both published the videos in question and obtained relevant profit for the purpose of promoting the Defendant’s books in accordance with the platform rules and relevant service agreement, and the Defendant possessed the ability to review and manage the said videos on the basis of the platform rules and management access. The Court also considered that under the circumstance of massive use of the Plaintiff’s image and synthesized and simulated voice, the Defendant should have predicted that the videos had the risk of infringement and conducted a reasonable examination of whether the said videos obtained the authorization or not; however, due to failure to perform the reviewing obligations as it should be, the Defendant should be jointly liable for the infringed videos with the promoter. The Plaintiff has the right to choose the Defendant to be liable for all the compensation, which was supported by the Court. Accordingly, the Court ordered the Defendant to “apologize to the Plaintiff, compensate the economic loss and the reasonable cost for maintaining the right to RMB 120,000, and rejected the Plaintiff’s other requests.”
It is reported that the first instance judgment has come into effect and the Defendant has performed its obligation under the said judgment.
This is a typical case regarding the infringement of personal rights; however, it is also quite special, as it does not only involve the protection of the right of image but also the legal protection of the right of voice in the context of AI synthesis technology. In fact, the relevant content can be found in its legal basis in the PRC Civil Code. In accordance with Article 1019.1 of the Civil Code, “No organization or individual may infringe upon any other person's right to image by vilifying, defacing, or forging by means of information technology or otherwise. Without the consent of a person entitled to an image, the image of such person shall not be produced, used, or made public, unless otherwise provided by law.” Further, Article 1023.2 of the Civil Code provides that “The relevant provisions on protection of right to image shall apply as reference to protection of voice of natural person.”
In this case, the court further specified through the actual case that “natural human voices synthesized using AI deep synthesis technology are identifiable if the general public or relevant professionals can recognize a specific natural person based on their tone, intonation, pronunciation style, etc., and should therefore be included in the scope of protection of that natural person's voice right.” This ruling not only helps deepen our understanding of the application of the above provisions of the Civil Code but also provides important reference rules for similar cases in the future.
It is also worth noting that as the AI-generated synthetic images and sounds become increasingly realistic, this court ruling also imposes higher due diligence obligations on businesses and platforms that utilize such technologies for promotional purposes. For businesses, this means they should review the sources of promotional materials within reasonable limits (e.g., whether they have obtained authorization from rights holders) and retain evidence of fulfilling their review obligations. For platforms, this means they should label the AI-generated content in accordance with relevant laws and regulations and establish complaint mechanisms to better address new types of infringement, such as the unauthorized use of AI to infringe on image right or voice right.
In a word, this case not only provides important reference value for the effective protection of image right and voice right in the context of AI-generated synthetic content but also offers useful guidance on how to regulate the development of AI-generated synthetic technology.