Editor's Note: AI-powered short dramas are increasingly caught up in disputes over "face theft" (unauthorized AI use of facial images). Cases range from the unauthorized use of celebrities' likenesses — such as Yi Yangqianxi and Yang Zi — in AI-generated productions to the recent takedown of the AI short drama Peach Blossom Hairpin for misusing private individuals' facial images. Together, these incidents point to a recurring disregard for legal boundaries. Two experts share their views with China Daily's Li Wei:

Digital shortcuts meet legal dead-ends

The "face theft" practices in AI short dramas already amount to infringement — and may well cross into illegality. Recent regulatory guidance states that using personal data for modelling, image generation, or scene creation must respect individuals' lawful rights and must not violate portrait, reputation, honor, privacy, or personal information rights, nor violate personal dignity through defamation or distortion. In this light, recreating someone's likeness with AI for short drama without consent not only risks infringing civil rights, but can also expose those involved to criminal liability, may involve offenses such as identity fraud, defamation, and the production of illicit content, while undermining order in the digital space and posing broader risks to public security.

The rise of "face theft" is driven by a convergence of technology, economics, and regulatory lag. Deepfake technologies and open-source tools have sharply lowered the barrier to producing realistic faces, enabling even non-specialists to generate convincing results with minimal input. At the same time, AI-generated likenesses can slash costs from millions of yuan to tens of thousands and compress timelines from months to days, incentivizing shortcuts on obtaining portrait rights authorization. Regulation, however, has failed to keep pace, leaving gaps that allow such practices to persist.

Zhang Linghan is the dean of the School of Artificial Intelligence and Law at the China University of Political Science and Law.

Turning ethical rules into enforceable safeguards

From a legislative perspective, a clearer and more enforceable framework is essential to protect rights holders. Where an AI-generated character is readily identifiable as a specific individual, infringement should be presumed. Producers should not be permitted to invoke "accidental resemblance" or "non-commercial use" as defenses, nor should platforms rely solely on copyright licensing agreements to claim they meet their review obligations. The burden of proof must be strengthened on both sides: producers should demonstrate that any likeness used is duly authorized, while platforms must show that effective filtering and review systems are in place. At the same time, the cost of infringement must outweigh the gains. Introducing punitive damages for deliberate "face theft," alongside requirements for both producers and platforms to deploy proactive detection and filtering mechanisms, would help turn legal rules into enforceable, operational safeguards.

Technological neutrality is not responsibility neutrality. Technology may run wild, but rights cannot be left unguarded. The rise of AI is reshaping creation while testing the limits of governance and ethics. The healthy growth of AI short dramas cannot depend on after-the-fact penalties alone; it requires a shared front line of safeguards built by platforms, creators, developers, and users alike. Only when innovation advances alongside accountability, and industry growth proceeds with rights protection, can AI truly serve and respect people.

Shen Xie is an associate professor at the College of Marine Culture and Law, Jimei University.

The views don't necessarily reflect those of China Daily.

If you have a specific expertise, or would like to share your thought about our stories, then send us your writings at opinion@chinadaily.com.cn, and comment@chinadaily.com.cn.