First case on AI face replacement before the Internet Court in Beijing

With the rapid development of artificial intelligence (“AI”) and deep synthesis algorithms, “face swapping”, “makeup swapping” and other applications are becoming more widespread and popular, while at the same time, the risk of infringement of the personal rights and interests of natural persons in the application of such AI technologies is increasingly coming into focus.

On June 20, 2024, the Beijing Internet Court issued two first-instance judgments in the Beijing AI face-swapping app infringement case. The court found that the service provider who used other people's videos to “swap faces” and then created templates violated the personal information rights and interests of others, even though it had hired a third-party company to provide “face-swapping” services for profit.

The plaintiffs in this case were short video models. Their images were used without permission by the defendant, an operator of the AI ​​face-swapping app, to create face-swapping templates for paid use. The plaintiffs claimed that this violated their rights and interests in their portraits and personal information and demanded an apology and compensation from the defendant.

The case focuses on: 1) whether the image templates used for AI face swapping qualify as identifiable portraits; and 2) whether the integration of a third-party AI face swapping technology into the defendant's app means that the plaintiff's facial information is not processed by the defendant, which could mean that the defendant has not violated the plaintiff's personal data rights and interests.

The court ruled that although the templates generated by the AI ​​replicated the plaintiffs' movements, physical appearance and other characteristics, the differences in facial features did not allow them to be attributed to a specific person. Therefore, the court found no infringement of image rights.

However, the plaintiffs’ video footage contained personal information, such as their facial features, which met the definition of “personal information” under the Law on the Protection of Personal Information of the People's Republic of China. The defendant's unauthorized collection and use of this video material to create face swap templates and technically replace the plaintiff's faces with them constituted processing of personal data. Although the defendant had engaged a third party to provide the AI ​​face swap technology, the defendant decided on the nature and extent of the data processing and should therefore be held liable.

The plaintiffs' video footage, although publicly available, was clearly marked as not licensed for paid software. The defendant's unauthorized commercial use constituted a violation of the plaintiffs' rights and interests. The applicable laws are designed to protect personal data through the right to information and the right to make decisions to prevent the disclosure and misuse of personal data. The defendant could not prove that it had obtained the plaintiffs' consent and therefore violated the plaintiffs' rights and interests in personal data.

For app operators in the AI ​​field, this case provides two important lessons: 1) Engaging a third party to process personal data does not relieve them of their own responsibility. 2) Providers of AI-based face-swapping services must still obtain the consent of the original rights holder to avoid infringing on the rights and interests in personal data, even if they have deleted the facial features of the original rights holder when creating the image templates.

If you need further information, please contact our TMC/IP team.