(Yicai) Feb. 13 -- Chinese social media platform Xiaohongshu said today that it will restrict the distribution of artificial intelligence-generated content that is not proactively labeled, after highly realistic videos created by ByteDance’s Seedance 2.0 model sparked widespread controversy.

The move follows multiple user complaints about AI-generated videos impersonating public figures and producing fake content, raising concerns about portrait rights and trust within the online community. The policy also comes as Chinese regulators step up scrutiny of untagged and misleading AI content.

In an announcement published on its official account today, Xiaohongshu said some AI-generated posts had failed to include active labels, infringing on the legitimate rights and interests of public figures and undermining user trust. The platform added that it will continue improving its ability to identify AI-generated content and encourage users to proactively label such posts, while restricting the distribution of content that lacks clear AI disclosure.

Recently, a large number of videos generated by ByteDance’s Seedance 2.0 have circulated widely online. The clips closely resemble real footage and highly replicate the likenesses of certain public figures or film and television intellectual properties, drawing significant attention while also triggering debate over alleged portrait rights violations.

On Feb. 10, the Seedance 2.0 operating team said in a statement that it is conducting urgent optimizations in response to public feedback and has suspended the function that allows users to generate videos using real images or videos as reference materials.

China’s cyberspace regulator has also intensified enforcement against AI misuse. Yesterday, the Cyberspace Administration said it had investigated over 13,420 accounts that used untagged AI-generated content to deceive and mislead the public, and removed more than 543,000 pieces of violating information.

The regulator will also launch a one-month special campaign targeting the misuse of AI technology to generate vulgar or violent content, spread false information, or engage in malicious traffic-driven hype, it added.

Editor: Emmi Laine