Unauthorized Face Scraping for AI Short Dramas Violates Rights in China
AI short drama producers in China are sourcing virtual character faces by scraping and fusing photos of ordinary social media users without consent, then selling these generated face packages on second-hand platforms for minimal cost. The practice violates privacy rights and intellectual property protections, with some platforms removing infringing content after investigation.
What happened
In China's rapidly expanding AI short drama market, a grey market has emerged where people's facial images are scraped from social media platforms without consent, fused using AI tools, and sold as virtual character assets to drama producers. Vendors openly advertise face fusion packages on second-hand platforms for as low as 19.8 yuan (approximately $2.80 USD), containing nearly 500 facial asset variations.
The source material for these AI-generated faces comes from screenshots of ordinary social media users, particularly women, extracted from video platforms and overseas social media by searching keywords like 'Chinese people,' 'beauty,' and 'fashion.' Vendors acknowledge the process completely bypasses authorization requirements, with per-image production costs estimated at only 0.02–0.04 yuan.
Reported impact
Female social media users have had their photographs publicly stolen and resold without consent or compensation. Some AI short dramas have been shut down after using poorly disguised stolen likenesses or distorting the faces of ordinary people in negative portrayals.
Actors in the legitimate short drama industry face employment displacement and unauthorized use of their likenesses, with some cases already ruled as copyright infringement when AI characters too closely resembled well-known performers.
Index analysis
The practice violates China's Civil Code provisions requiring consent before creating, publicizing, or using someone's likeness, and prohibits fabricating likenesses using information technology. Unauthorized large-scale collection and provision of facial data—a sensitive personal information category—may trigger criminal liability under laws protecting citizen privacy.
The incident reflects inadequate governance and oversight in the AI-generated content sector, where facial recognition and synthesis technology is deployed at scale without adequate consent mechanisms or content moderation systems.
Governance controls
A South China Media reporter deleted collected materials and reported incidents to platforms and relevant authorities. Some short dramas using stolen or defamatory AI faces have been removed from circulation.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.