TOPICS 

    Subscribe to our newsletter

     By signing up, you agree to our Terms Of Use.

    FOLLOW US

    • About Us
    • |
    • Contribute
    • |
    • Contact Us
    • |
    • Sitemap
    封面
    NEWS

    China Slams AI-Generated Memes for ‘Desecrating’ Traditional Culture

    Netizens are using AI video tools to spoof all manner of Chinese classics. The authorities are not amused.

    The dawn of AI video generation technology has produced one notable effect in China this year: a deluge of clips spoofing classical works of Chinese literature. And the authorities are not happy about it.

    China’s social platforms have been flooded with videos reimagining famous scenes from Chinese classics — from the Monkey King speeding through his “Journey to the West” on a motorcycle, to the ancient strategist Zhuge Liang casually munching on a durian amid a fierce battle.

    These light-hearted memes are proving highly popular with China’s internet users, receiving thousands of shares. But the trend is enraging the government, which has slammed the videos as a “desecration” of the country’s cultural heritage.

    The National Radio and Television Administration (NRTA) issued a statement on Dec. 7 calling for provincial authorities and internet platforms to clamp down on AI-generated content that “wickedly modifies” traditional Chinese culture.

    “These videos, created purely to chase traffic, blatantly desecrate classic intellectual property without restraint, undermine traditional cultural values, contradict the core essence of the original works, and may constitute copyright infringement,” the notice stated.

    Video generation tools have surged in popularity in China since early 2024, when OpenAI’s preview of its new Sora tool shocked the Chinese tech industry and sparked a rush to produce domestic versions of the software.

    Many of China’s tech giants have now released their own AI video tools, with Kuaishou debuting Kling AI in June and ByteDance adding video-generation features to its Jimeng AI in November.

    Netizens quickly realized the tools were ideal for producing spoof memes. Users can upload reference images from TV series based on Chinese classics and add text prompts to generate videos, which can range from a few seconds to a few minutes long.

    Some platforms can even simulate the original voices of the characters, synchronize lip movements, and provide voiceovers based on the user’s instructions. In recent weeks, a string of AI-altered clips have gone viral on Chinese video platforms including Douyin, Kuaishou, and Bilibili.

    The series “‘Romance of the Three Kingdoms’ in AI” — which features Iron Man being executed after a battle and Alibaba founder Jack Ma being crowned emperor — has already garnered about 350,000 views on Bilibili.

    But the clips have also generated pushback from regulators, who have accused some videos of violating the law.

    In September, China proposed new regulations mandating that all AI-generated content must be clearly labeled with watermarks and embedded metadata, as part of an effort to stem a surge in AI-related fraud.

    Chinese social platforms have been left to enforce this policy themselves, with most requiring creators to manually check a box to confirm their content is AI generated before they publish it.

    Some of the viral spoof videos do display the required watermarks, but others do not.

    There have been calls for platforms to become more proactive in detecting AI content and adding labels themselves, but experts say that companies would need to invest more in their detection capabilities to do this.

    “We should impose stricter compliance requirements on generative AI service providers,” said Ma Ce, an attorney specializing in AI law at Zhejiang Kinding Law Firm.

    Ma added that some AI-generated videos may be violating Chinese intellectual property regulations.

    “The use of generative AI should respect others’ legitimate rights, including personality and intellectual property rights,” Ma said. “Altering traditional cultural classics should not devolve into defamation or unauthorized manipulation of characters and audiovisual materials.”

    Douyin, Kuaishou, and Bilibili have yet to publicly respond to the NRTA’s notice, but some viral AI-generated clips have been removed from the platforms.

    Some AI video tools have also taken steps to prevent the use of stills from classic films and TV shows, according to domestic media reports.

    The NRTA notice instructed provincial authorities to make sure that video platforms in their jurisdictions inspected and removed AI-generated videos based on classic films and TV dramas by Dec. 10.

    China has a long history of spoofing classic books, films, and TV series. A few years ago, a subculture known as lalangpei — or “forced pairing” — took off on the Chinese internet, which saw creators stage romances between unlikely combinations of popular characters.

    One famous lalangpei, for example, brought together Lord Voldemort from the “Harry Potter” series and Lin Daiyu from the classic novel “Dream of the Red Chamber.”

    A related subculture known as guichu, which was also hugely popular years ago, saw creators make mashup videos in which clips of people speaking are cut together to create satirical songs or humorous videos.

    One popular guichu video, which cut together clips of the ancient general Cao Cao from “Romance of the Three Kingdoms” and Benedict Cumberbatch in “Doctor Strange,” amassed over 6 million views.

    But the guichu trend began to decline in 2018 after the NRTA issued a ban on videos cutting together, mixing, recaptioning, and redubbing original works, as well as parodies of “classic works.”

    (Header image: From Douyin, re-edit by Sixth Tone)