AI-Related Litigation in China

Introduction

China has developed a rapidly growing body of case law addressing legal questions arising from artificial intelligence. Chinese courts, particularly the Internet Courts established in Hangzhou, Beijing, and Guangzhou, and the specialised intellectual property courts, have adjudicated disputes involving AI-generated content, facial recognition technology, AI-related tort liability, and data scraping. The case law, though still developing, provides important guidance on how Chinese law applies to AI-related disputes in the absence of a comprehensive AI-specific legislative framework.

Tencent v Shanghai Yingxun: AI-Generated Content

Tencent Technology (Shenzhen) Co., Ltd. v Shanghai Yingxun Technology Co., Ltd., decided by the Nanshan District People’s Court of Shenzhen in 2019, was the first Chinese case to address copyright in AI-generated content. Tencent’s proprietary AI system, Dreamwriter, had generated a financial news article on stock market data. Shanghai Yingxun reproduced the article on its own platform without authorisation. Tencent sued for copyright infringement.

The court held that the Dreamwriter-generated article constituted a copyright-protected work under Chinese law. The court reasoned that the article reflected the “intellectual achievement” of Tencent’s human team, who had designed the AI algorithm, selected the data inputs, and set the creative parameters for article generation. The court distinguished the case from purely machine-generated content by emphasising the human contribution to the creative process.

The Tencent v Shanghai Yingxun decision established that AI-generated content may be protected by copyright where the human contribution to the creative process is sufficient. The court did not address the question of whether content generated entirely by AI without human creative input could be protected, leaving open the most difficult question in AI copyright law.

Beijing Feilin v Baidu: AI Copyright

Beijing Feilin Law Firm v Baidu Corporation, decided by the Beijing Internet Court in 2020, addressed the related question of authorship in AI-assisted works. Feilin used an AI text-generation tool to produce a market analysis report and registered copyright in the report. Baidu reproduced the report on its platform. Feilin sued for infringement, and Baidu challenged the validity of the copyright registration on the ground that the report had been generated by AI.

The court distinguished the case from Tencent v Shanghai Yingxun. In Feilin v Baidu, the court found that the human contribution to the AI-generated output was insufficient to establish authorship. The plaintiff had merely entered keywords and selected output options, without exercising sufficient creative control over the final product. The court held that the AI-generated text did not qualify as a work protectable under the Copyright Law.

The two decisions create a spectrum of protection: where human creative input is substantial, AI-generated content may be protected (Tencent); where human input is minimal, the output falls into the public domain (Feilin). The lower courts have not articulated a clear standard for distinguishing the two categories, leaving uncertainty for AI content creators and users.

Facial Recognition Privacy Cases

The Guo Bing v Hangzhou Safari Park case, decided by the Hangzhou Intermediate People’s Court in 2021, was China’s first facial recognition privacy case. Guo Bing, a consumer, had purchased an annual pass to Hangzhou Safari Park that required fingerprint recognition for entry. When the park upgraded its system to facial recognition and directed Guo to register his facial data, he refused and sued for violation of his personal information rights.

The court held that the park’s collection of facial recognition data without the plaintiff’s explicit consent violated the Personal Information Protection Law (PIPL) and constituted infringement of personal information rights. The court ordered the park to delete the plaintiff’s facial data and to pay damages. The decision was widely cited as establishing the principle that biometric data collection requires specific, informed consent independent of general terms of service.

Following Guo Bing, Chinese courts have adjudicated a growing number of facial recognition disputes. In Liu v Shenzhen Metro, the Shenzhen court held that a metro operator’s mandatory facial recognition requirement for weekly pass holders violated PIPL consent requirements. In Wang v Alipay, the Beijing court found that Alipay’s default facial recognition settings constituted inadequate consent.

AI Tort Liability

Chinese courts have begun to address tort liability for AI-caused harm under the Civil Code and the Product Quality Law. The Civil Code’s general tort provisions (Articles 1165-1177) apply to AI-related harm through principles of fault-based and strict liability.

In Chen v Didi Chuxing, the Beijing court addressed liability for an AI-powered ride-hailing platform’s algorithmic pricing. The plaintiff alleged that Didi’s algorithm charged her a higher price than other users for the same route on the same day (the “big data price discrimination” problem). The court held that the platform’s use of algorithmic price differentiation based on user profiling constituted a breach of the platform’s duty of good faith under the Civil Code and ordered compensation.

Li v Xiaomi Corporation, a product liability case, addressed harm caused by a smart home AI device. The plaintiff’s Xiaomi smart speaker issued an incorrect voice command that triggered a connected appliance to start a fire, causing property damage. The court applied the Product Quality Law’s strict liability provisions, holding Xiaomi liable for the defective design of the AI system that failed to verify the command before execution.

Data Scraping Disputes

Data scraping disputes have emerged as a significant category of AI-related litigation. In ByteDance v Sina Weibo, ByteDance sought an injunction against Sina Weibo’s scraping of Toutiao content for AI training purposes. The court framed the question as unauthorised data access under the Anti-Unfair Competition Law, holding that systematic scraping of a competitor’s data constituted unfair competition.

In Tencent v Beijing Huaqiao Technology, the court addressed the use of publicly available data for AI model training. Huaqiao Technology had scraped publicly accessible user posts from Tencent’s social media platform to train its natural language processing AI. The court held that the scraping of publicly available data was not per se unlawful but became unlawful when the scraping volume, frequency, or method imposed an unreasonable burden on the platform’s server capacity or circumvented technical protection measures.

Conclusion

Chinese AI litigation is developing through case-by-case adjudication in the absence of a comprehensive AI law. The Internet Courts, IP courts, and ordinary courts have produced significant decisions on AI-generated content copyright, facial recognition data rights, AI product liability, and data scraping. The decisions reveal a pattern of Chinese courts applying existing legal frameworks — copyright law, the PIPL, the Anti-Unfair Competition Law, and the Civil Code — to AI disputes, adapting established doctrines to novel questions. A comprehensive AI Law, currently under legislative consideration, may codify and extend the principles established in the case law.