China: Courts and AI face off

Courts in China are increasingly rejecting ‘algorithmic neutrality’ as a defence in AI cases—and businesses need to adapt, say Xiaojun Guo and Hongxia Wu of CCPIT Patent and Trademark Law Office.

The rise of artificial intelligence (AI)-driven technologies, such as deep synthesis algorithms, has revolutionised content creation, enabling applications like one-click “face-swapping” and “makeupswapping”.

While these innovations enrich digital experiences, they also raise significant legal concerns, particularly in copyright and personal information protection.

Recent cases in China highlight how courts are addressing these challenges, offering critical guidance for balancing technological innovation with legal accountability.

Case 1: Copyright infringement in AI face-swapped videos

In April, the Supreme People’s Court released a landmark copyright case made by the Shanghai Jiading District People’s Court (Case No. [2024] Hu 0114 Min Chu 1326). The plaintiff, Chen, had posted 13 original short videos on Douyin (TikTok), featuring women in traditional costumes.

A tech company developed a miniprogram that allowed users to replace the faces in Chen’s videos with their own via AI synthesis.

The modified videos retained the original scenes, camera angles, character modelling, and movements, differing only in facial features. Users could access the service by watching ads or purchasing memberships, generating revenue for the company.

Court’s analysis

1. Originality of Chen’s work: The court affirmed that Chen’s videos constituted audiovisual works under copyright law. Their creative selection and arrangement of the content arrangement, scene selection, shooting angle, etc, reflected sufficient originality.

2. Substantial similarity: The AI-modified videos replicated the core creative elements of Chen’s work, differing only in facial details. This constituted unauthorised reproduction.

3. Rejection of “technical neutrality” defence: The court rejected the defendant’s claim of passive algorithmic processing, noting that the company actively promoted “AI face-swapping” as a commercial feature and profited from it.

4. Fair use inapplicable: The modifications were neither transformative adaptations nor covered under permissible uses like parody or education.

Outcome: The defendant removed the infringing content, registered its algorithms, and accepted judicial guidance on ethical AI deployment. Chen withdrew claims for an apology, and the court ordered only monetary compensation.

This case clarifies that AI-assisted modifications do not circumvent copyright obligations, even if partially altered.

Case 2: Personal information rights in AI-generated templates

In June 2024, the Beijing Internet Court addressed a dual claim involving portrait rights and personal information infringement. Two models discovered their likenesses had been used in an AI face-swapping app’s video templates.

While their faces were replaced via deep synthesis, their dress-ups, hairstyles, clothing, movements, lights, and camera switching remained intact. The plaintiffs argued that their portrait rights and personal information rights were violated.

Court’s analysis

1. No portrait right infringement: The replaced facial features made the plaintiffs unrecognisable, negating claims under China’s portrait rights laws.

2. Personal information violation: The app had obtained the plaintiffs’ biometric data such as facial features, and processed the facial features, without consent. Even anonymised, such data qualifies as personal information under China’s Personal Information Protection Law.

Outcome: The defendant committed illegal processing of the plaintiff’s personal information, using their personal data in commerce, even if anonymised, shall be liable for its conduct.

Legal implications for AI innovation

These court rulings establish critical precedents for AI governance:

1. Copyright protection in the AI era

Originality threshold: Courts recognise AI-modified content as infringing if it retains the original work’s creative essence (eg, scene arrangement, camera angles).

Service provider liability:

Companies leveraging AI tools bear responsibility for ensuring content legality. “Technical neutrality” defences fail if the technology is marketed for infringing uses. Fair use limits: Minor modifications (eg, face-swapping) lack transformative value and do not qualify as fair use.

2. Privacy and data compliance

Even anonymised data used for commercial purposes mandates user authorisation, aligning with global trends like the EU’s GDPR.

3. Proactive compliance strategies

To mitigate risks, businesses should: Audit training data: Ensure copyrighted materials are licensed or in the public domain.

Implement consent frameworks: Obtain explicit user permissions for data collection and processing in advance.

Register algorithms: Comply with regulations like China’s Deep Synthesis Management Provisions, which mandate transparency for AI tools.

Conclusion: Striking the balance

These cases mark a pivotal step in defining legal accountability for AI-driven content. As courts increasingly reject an “algorithmic neutrality” defence, businesses must integrate rights clearance and privacy safeguards into AI development.

The rulings strike a delicate balance—fostering innovation while ensuring creators and individuals retain control over their works and identities in the digital realm.

China has been actively exploring the establishment of a dedicated legislative system for AI. This includes advancing the formulation and implementation of regulatory documents like the Interim Measures for the Management of Generative AI Services.

The legislative initiative not only aims to provide institutional safeguards for AI technological innovation, but also prioritises the construction of risk prevention mechanisms across multiple dimensions, including algorithm governance, data compliance, and ethical review.

By striving to achieve a dynamic equilibrium between technological advancement and rights protection, this approach seeks to contribute Chinese solutions to the development of global AI governance paradigms.

Takeaways

• Rulings establish critical precedents
• Liability and fair use limits addressed
• Safeguards in AI development are vital

Xiaojun Guo is a patent attorney of CCPIT Patent and Trademark Law Office and the secretary general of AIPPI China.

Hongxia Wu is the deputy director of the Trademark and Copyright Litigation Department of CCPIT Patent and Trademark Law Office.

They can be contacted at: guoxj@ccpit-patent.com.cn or wuhx@ccpit-patent.com.cn