TLDRocket
Sign in

Prove You're Human

ChinaTalk Zilan Qian

Chinese illustrators are livestreaming themselves drawing, just to prove a human made the art. AI trained on their work, and now flagging tools keep mistaking their art for machine output.

Based on reporting by ChinaTalk, Zilan Qian — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

In China's illustrator community, a strange ritual has taken hold: artists betting each other that they can redraw their own work live, on camera, just to prove no algorithm touched it. The trigger was a labelling law that came into force in September 2025, requiring platforms to auto-flag suspected AI-generated content. One artist, going by "Angel from Mercury," found a commissioned piece flagged as "may contain AI" on Xiaohongshu. She signed a bet with another illustrator and streamed herself completing the illustration from scratch. She finished. The accuser wasn't convinced anyway.

That standoff, which happened on October 29, 2025 and went unresolved, sits inside a longer fight. Back in December 2023, four illustrators sued Xiaohongshu at the Beijing Internet Court, in what's described as China's first case testing whether training an AI model counts as copyright infringement. Their target was Trik AI, a tool Xiaohongshu launched in April 2023 that let users upload a photo and get artwork back. The illustrators said the output echoed their own brushwork closely enough to be recognizable. Xiaohongshu called it fair use. No verdict has been published, though the company quietly pulled Trik AI once the case drew attention, while similar generation tools kept launching elsewhere.

The pattern isn't accidental, according to legal scholars cited in the piece: China's regulatory posture has been described as "moderate leniency," tight rules on what comes out of AI systems, loose ones on what goes into them. That means illustrators have little recourse when their work becomes training data, but plenty of exposure to a labelling regime that, in practice, can't reliably tell their work apart from a machine's. Since 2022 the illustrator community has run its own parallel enforcement, bet-on agreements and live drawing sessions meant to out suspected AI use, because using AI-generated art carries real stigma given how the training data was obtained in the first place.

Meanwhile the job-market backdrop keeps darkening. China Economic Weekly flagged illustrators back in May 2023 as among the first workers laid off because of AI, at a time when tools like Stable Diffusion and Midjourney were already good enough to erode parts of the workflow. Universities have followed suit: Chinese schools eliminated 12,000 majors between 2021 and 2025, most of them in humanities and arts, and the Communication University of China cut 16 programs this year alone, including illustration, long one of its flagship majors. Southern Daily reported that by 2023 art-design jobs at game companies, especially smaller studios, were already disappearing.

Not everyone treats this as pure loss. The state has been promoting AIGC creation since 2021 through campaigns and competitions, and a community of self-described AIGC artists, some with no formal art background, has grown around the idea that AI is a tool rather than a threat. Filmmaker Jia Zhangke made an AI short film; ink painter Jia Hongjun, a professor at the Central Academy of Fine Arts, mounted an AI-generated exhibition. Whether that optimism holds up against the layoffs and the livestreamed loyalty tests is the question the piece leaves open.

My take — AI-written commentary, not fact-checked reporting

The livestream-to-prove-you're-human routine is a symptom, not a fix. It puts the burden of proof on the exact people whose work got fed into the machine without asking, while the platforms that built leaky labelling systems face none of it. Anyone who thinks stigmatizing "AI-looking" work protects human creators should watch what it actually does here: illustrators start second-guessing their own style so they don't get mistaken for the thing they're trying to prove they aren't. That's not a defense of humanity, that's humans editing themselves into something blander to dodge suspicion.

Read more about this at: ChinaTalk

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.