China Bans 'Magic Editing' of Revolutionary Heroes in AI Video Crackdown

Beijing’s media regulator has launched a month-long effort to curb AI-generated videos that make classic Chinese works seem vulgar. Now, platforms must check all AI content before it goes online.

On January 1, China’s National Radio and Television Administration launched a campaign against AI-generated videos featuring revolutionary heroes in comedic roles. The campaign targets AI parodies of Journey to the West, Romance of the Three Kingdoms, and Communist Party-themed movies. Channel NewsAsia reports that regulators are looking for content that “distorts, parodies or vulgarizes” these important cultural works.

This is a big change from September, when platforms only needed to add watermarks to AI content. A recent Xinhua report says Chinese authorities now require all online AI-generated content to be clearly labeled, starting September 1. By then, platforms like Bilibili and Xiaohongshu had already established systems to check metadata. Li Hongda, chief engineer at Shanghai Information Security Testing Evaluation and Certification Center, explained, "AI platforms and service providers must now label AIGC so that people can tell when something was created by AI and where it came from."

Labeling alone was insufficient. The new campaign now targets what regulators refer to as "magic editing," which involves AI filters and effects that allow creators to reimagine classic characters. For example, a deepfake of the Monkey King singing modern pop songs is now prohibited. Revolutionary soldiers depicted with face-swap filters will also be subject to administrative penalties. Bilibili has taken steps to show it is following the rules. According to Yicai Global, Bilibili has notified users about the new regulations and its AI-labeling feature multiple times through announcements, site messages, pop-ups, and explanatory videos, as stated by Luan Shuai, the platform’s head of content production. ’ The campaign sets clear boundaries. The Epoch Times’ review of the regulations says that authorities are “prioritizing the removal of AI-altered clips derived from the Chinese regime’s historical dramas, revolutionary films, and content portraying ‘heroes.’” This is not about deepfake detection or fighting misinformation. Instead, it is about controlling how AI tools are used with political stories and symbols.

For creators of "micro-dramas," which are short, serialized videos popular on Chinese social platforms, these rules are a big limitation. These videos often use AI effects to make the most of small budgets and go viral. No creators were interviewed, but it is clear that the filters they use now carry regulatory risks.

The systems needed for this crackdown were already in place. Since September, major platforms have checked metadata and added watermarks to AI content. MiniMax's Hailuo video generator has followed labeling rules from the start. The Global Times says Douyin is now limiting some accounts, including those run by government media or focused on political, social, economic, legal, or medical topics, from making commercial deals involving content publication. Bilibili has not shared details about how it reviews AI-generated content before it goes online. It is still unclear how these rules will be enforced.

How will officials distinguish between creative reinterpretation and vulgarization? The NRTA's month-long campaign ends on January 31, but the requirement for pre-publication review of all AI content suggests this is not just a temporary campaign but a new regulatory reality in China.