In China, as artificial intelligence (AI) micro-dramas sweep the video content market, the problem of ‘face theft’ has emerged as a social issue. Reports alleging that faces were used without consent are mounting, and there are growing claims that human faces have become the ‘raw material’ of the AI content industry.

According to Chinese outlets Jaeil Jaegyeong and Pengpai News on the 6th, an internet user named ‘Baichai Hanfu Makeup (Baichai)’ posted at the end of last month that her photos and face had been stolen for the period-style micro-drama <Dohwajam>, writing: “(The producer) used photos I had posted before without permission and edited them with AI. The clothing, accessories, and makeup of the drama character are exactly the same as mine in real life,” and “In the drama, I am portrayed as greedy and lewd. It is truly absurd.”

After Baichai went public, complaints poured in from both ordinary people and celebrities saying their faces had been used in AI dramas without permission. The female model Chi Hai also said her face was taken for <Dohwajam> and that being depicted in a malicious way hurt her modeling career.

The agency of idol-turned star Yi Yangqianxi (26) issued a statement on the 5th saying that multiple dramas on Hongguo including <Midnight Bus: Her Schemes Are Vicious> and <You Are Lying to Reincarnate Me? Fine, Do Not Regret It> had appropriated his face. The agency stated, “Yi Yangqianxi has never appeared in the related dramas, nor has he granted AI the right to use his image.”

Hongguo said it would remove the problematic dramas from the platform and suspend creators from uploading content for 15 days.

The Actors Committee of the China Broadcasting and Exhibition Organization Federation issued a statement on the 2nd saying, “Unauthorized use of the image and voice of actors infringes portrait rights and voice rights,” and pledged support for actors pursuing legal action.

Chinese media reacted that ‘what was bound to happen has arrived’. Jaeil Jaegyeong wrote, “AI micro-dramas are currently in a phase of rapid expansion. With advantages such as low production costs, short cycles, and clear monetization pathways, the field has quickly drawn in many content producers,” adding, “But at the same time, the gray-zone phenomenon where stealing the faces of stars or ordinary people blurs the boundaries of legal rights is becoming an ‘unspoken rule’ of the industry.”

Produced by an animation and drama studio in Chengdu, <Dohwajam> is a hit that, as of the end of last month, had amassed over 40 million cumulative views. Period dramas typically incur high costs for costumes, makeup, and sets, but by using AI production this series sharply cut costs and released installments quickly, and it had been cited as an example of ‘innovation’ in the industry.

A micro-drama typically runs 1 to 5 minutes per episode, and some are under 1 minute, with a total of 40 to 60 episodes. Platforms offer the first 5 to 6 episodes free, then prompt viewers to pay for subsequent installments. As Chinese AI companies such as SenseTime released programs that generate video and music from scripts so a drama can be made in one go, the barrier to entry for producing AI dramas has fallen sharply. Not only established production houses but also one-person companies have jumped in.

According to the South China Morning Post (SCMP) in Hong Kong, China's micro-drama market this year is estimated at about 90 billion yuan (about 18.9 trillion won), up 78% from the previous year. Chinese authorities are also encouraging AI content production to advance AI technology and to develop overseas content markets. There is also a perception that, because micro-dramas have no overseas competitor on par with China, they could enable China to sweep the world, like K-pop or Japanese animation.

However, as face theft comes to the fore, the flip side of AI-drama innovation is drawing scrutiny. Columnist Jang Tong wrote in a Pengpai News column, “It has become an undeniable reality that, in the AI content industry chain, the faces of ordinary people have become a kind of ‘raw material’ that can be collected, processed, and monetized at low cost,” adding, “This process does not require personal consent and carries virtually no legal risk.”

For non-celebrities, it is hard to bring individual lawsuits, platforms quickly take down the dramas, and producers use AI to re-edit images so no evidence remains. Under current law, the standard for judging whether portrait rights have been infringed is ‘can others recognize the face’; if AI blends diverse facial features to create a new face, it becomes harder to have ‘theft’ recognized.

Jang Tong argued that to prevent rights infringements, rather than focusing on the ‘identifiability’ of AI-generated content, it is necessary to enable tracing of the source materials collected in the process of producing AI content and, through that, to apply a comprehensive standard of whether the ‘content production process was lawful’.