Get Started

When AI Dramas Get Pirated: China’s First Criminal Copyright Case

Written by Gilbert I. Kangdra BBA. | Published 

November 3, 2026

AI-generated short dramas are one of the fastest-growing formats on Chinese video platforms. In May 2026 a court in Guangzhou handled a question the format had made unavoidable: if a drama was made largely by AI, is it protected by copyright, and can copying it be a crime? The court said yes to both.

The case

The Huangpu District People’s Court in Guangdong delivered what reports describe as the Chinese mainland’s first criminal verdict over copyright infringement of AI-generated short dramas. According to Dim Sum Daily, two individuals received suspended prison terms for illegally recording and reselling more than 1,700 productions.

The court found that the pair used screen-recording software to copy the dramas in bulk. They then sold them at very low prices on second-hand online platforms. A Rouse analysis republished on Lexology says the volume was far above the 500-work threshold for criminal prosecution, which is why the procuratorate brought a public prosecution rather than leaving the matter to a civil claim.

The dramas were produced with AI tools developed by a gaming company. Rouse reports that they had grown into a series on short-video platforms with a large audience. Rouse also stresses that the evidence showed the dramas were not the product of a simple one-click process. That detail turned out to matter for the legal question.

The legal question: is AI output a “work”?

The central ruling is that the AI-generated dramas qualified as works under Chinese copyright law. The court’s reasoning, as summarized by Rouse, is that users who create dramas with AI can receive copyright protection where the result is original. Protection is not automatic simply because AI was involved.

This follows a line of Chinese lower-court decisions that have pulled in different directions:

  • In November 2023 the Beijing Internet Court recognized copyright in an AI-generated image, in a widely noted first-instance ruling.
  • In March 2025 a court in Zhangjiagang denied protection to AI-generated pictures, the first Chinese case to do so according to King & Wood Mallesons.
  • In September 2025 the Beijing Internet Court announced an upheld decision saying copyright can exist in AI images, but only if the author can show creative effort and submit evidence of their prompts and their selection and editing process.

Taken together, the courts are not asking whether AI was used. They are asking how much human creative input shaped the result, and whether the creator can prove it.

A civil follow-up in Wuhan

On September 23, 2026, the Jiang’an District People’s Court in Wuhan announced what it called the first case in Hubei Province involving infringement of an AI-assisted short drama. National Law Review reports that the court held the drama to be a protected audiovisual work. It reasoned that although the visuals and voiceovers were generated mainly by AI tools, they were not the mechanical output of a single automated process.

The court also took AI token costs into account when calculating damages, which is an unusual step. A separate report cautions that this single ruling does not establish a nationwide rule on token costs.

The court gave creators practical advice. They should keep evidence of their creative process, including scripts, prompt drafts, generation records, original project files, and proof of first publication.

A related problem: faces and voices

Copyright is only one part of the legal risk around AI dramas. China Daily (Hong Kong edition) reports that several celebrities, including Yi Yangqianxi and Yang Zi, said they had never authorized the use of their images or voices in AI dramas. In the first quarter of the year, the platform Hongguo removed 1,718 animated short dramas for policy violations, and 670 of them were suspected of using people’s facial information without consent.

The same report describes a March court decision in favor of an actress whose likeness was mapped onto a character in a 90-minute drama using AI face-swapping. The judge said that even a slightly altered AI face can infringe if the public can recognize the person and no permission was given.

Platform liability

The short-drama cases build on earlier rulings about AI services themselves. In February 2024 the Guangzhou Internet Court found a generative AI provider liable for infringing the Ultraman series. Linklaters summarizes the ruling as finding infringement of both reproduction and adaptation rights, because the service generated images identical or substantially similar to the originals.

What this means

For creators. Protection depends on documented human creative input. Prompts, drafts, edits, and project files are evidence, so keeping them is cheap insurance.

For pirates and resellers. Screen-recording and reselling AI dramas can be treated like copying any other audiovisual work, including criminally when the scale is large.

For AI users generally. Rouse’s summary makes the point directly: using AI to create content does not guarantee ownership, and using it does not remove liability for copying someone else’s work.

Limits of the evidence

These are lower-court decisions. China’s courts do not treat them as binding precedent in the way common-law courts do, and judges in other regions have reached different conclusions on AI works. Reports also differ on some details of the Guangzhou sentence, so anyone relying on the specifics should check the court’s own published judgment. Finally, the Wuhan ruling on token costs is one court’s approach, not settled national practice.

Sources

 

© 2026 Tranfan Law Office. All Rights Reserved.

Recent News