A top concern for businesses in China is protecting your intellectual property, in an environment where data theft is common. Companies in China may copy data right off your site and host it on their own. A typical tactic used by thieves is to look for where data has somehow been publicly exposed, for example by employees in chat, APIs, or ungated content, then download the content and post it to their own platform. You can beat these companies in court, but most businesses are simply victims to do so because they don’t develop a strategy that takes into consideration the law in China. This article will explain common loopholes used by data thieves in China, and how to enforce your rights in your valuable IP rights. By arming yourself with this knowledge, you can be in a good position to beat IP thieves in Chinese courts.

Contents

• How Data Asset Claims Are Enforced
• Data Asset Claims Cases
• How to Protect Data Assets Under Chinese Law
• Unfair Competition Issues Related to Data Use in AI Training

How Data Asset Claims Are Enforced

In most jurisdictions, copied data would usually be treated as intellectual property, but in China these claims are currently usually brought as unfair competition, since the IP route is usually not viable. That means enforcement will be challenging, for a few reasons. Firstly, courts show a preference for applying judicial doctrine to these novel data protection disputes. The Unfair Competition Act’s rules are extremely broad; internet law rarely finds application.

Under these rules, courts impose unfair-competition liability when both unfair data acquisition and unfair data use are alleged,[1] though some recent cases consider unfair use alone sufficient. User consent is necessary to establish liability, and consent may amount to fair use. Courts assess harm through three factors: competitor interests, consumer interests, and market integrity.

Chinese courts do not necessarily require showing substitution in the markets to establish there is a competitor interest. “Consumer interest” means consumers’ personal data protection. Data and cyber security are increasingly less relevant to market integrity analysis. Overall, Chinese courts are adopting a more flexible, nuanced approach to conduct, consent, and harm, with personal data security and user consent strongly influencing outcomes.

Data Asset Claims Cases

Many businesses assume that if they control the site, they own the data, but this has turned out to be false and has resulted in lawsuits throughout China. In the judicial precedents, Chinese courts do not inquire into who hosts the data, and instead look at who creates it, how it is obtained, and whether its use is fair. Depending on the type of data, claims have been brought as unfair competition, copyright, or trade secrets.

Unfair Competition Cases

In Beijing Case 73-cv-1011 (2021), Company A ran a short form video platform aggregating user-uploaded videos, registration data, and comments under a EULA reserving all rights.[2] Company B scraped videos and user data, then reproduced and distributed them on its own app. The court found that some videos met the originality requirement for copyright protection, while others qualified as cinematographic works despite having little originality. The court determined that Company A owned no copyright in the videos or datasets but had protectable rights in profits derived from the data collection.

The court held that, although some platform content was copyrightable, the mere collection and categorization of short videos lacked sufficient originality for compilation copyright, so Company A could not claim copyright in the videos or content. Still, its substantial efforts to attract users and build commercially valuable datasets deserved protection as a legitimate economic interest. Even without blocking Company A’s services, Company B’s unauthorized scraping and reproduction functionally substituted Company A’s products and services, harming consumer interests and market integrity, and therefore violated good faith and fair dealing principles.

The court found Company B liable for unfair competition, ordered damages, and required publishing a newspaper notice to reduce harm. The 5 million Yuan damages award was upheld on appeal.

The case shows that platforms cannot automatically claim copyright in user-uploaded content in China. Courts examine the content’s original party and reject compilation copyright claims based only on generic categories such as “video,” “streaming,” or “music.” Compilation copyright requires originality in selection and arrangement of the assets. Under China Unfair Competition Act § 12, web scraping is not interference with another business’s products or services unless as a result is, those products or services cannot operate normally.[1]

Copyright Litigation

In Shanghai Case No. 73-cv-490 (2024), the defendant circumvented the plaintiff’s login system, access controls, and protocols to scrape user posts and other data, provided from a single API.[3] The plaintiff alleged copyright infringement in the user posts, and unfair competition that harmed its data-driven business model. The defendant responded that the data is not copyrightable work and that the platform’s form contract is not sufficient to assert copyright.

The court rejected the copyright claim, holding that only posts with originality of certain types are copyrightable. The users retain ownership, and the platform has only limited protectable rights under the user agreement or EULA, specifically, to maintain data security and stop large scale piracy. Although the defendant retrieved posts, there was no evidence of modification, republication, or infringement of any individual work. The court also held that without explicit user authorization, the platform could not seek punitive damages for copyright infringement.

It nonetheless found that the platform had a legitimate competitive interest in its account data and posts and held the defendant liable for unfair competition.

The court held that platforms’ rights in their content are limited to protectable rights under the EULA, including the rights to take action against large scale piracy. Platforms do not by their nature have copyright in individual posts or the right to seek punitive damages without user authorization.

Trade Secret Cases

In Shanghai Case 02-cv-11028 (2023), the plaintiff publicly shared its ex-factory prices for several years in WeChat groups and by phone, creating a large dataset of essential price data.[4] The defendant, a steel industry information provider, independently collected prices from the plaintiff and other companies to create and publish price indexes on its website and mobile app, a data service. After negotiations over data use failed, the plaintiff sued, alleging that the defendant’s unauthorized collection and processing of protected price data constituted trade secret misappropriation, along with false advertising and passing off.

The court held that ex-factory prices were public market information and that the plaintiff’s WeChat disclosures without confidentiality requirements made the data public, so it was not “not generally known to the public,” therefore not a trade secret. The court further emphasized that data free from national secret, personal privacy, and confidentiality restrictions should circulate freely; the law will not encourage creating data moats. It found that the defendant lawfully obtained the data through legitimate sources and methods without misrepresentation or fraud. The court therefore dismissed the plaintiff’s trade secret and other infringement claims.

How to Protect Data Assets Under Chinese Law

Enforcement Options

The cases above show three causes of action are available for data asset protection: copyright, trade secret, and unfair competition claims. Each has advantages and disadvantages depending on the circumstances.

• An unfair competition claim is flexible and broadly applicable, protecting data assets otherwise left unprotected. It does not require clearly established ownership and may cover merely collected or structured datasets, including those of public data. However, it’s subjective with no bright line rules. Instead, courts weigh competitors’ interest, conduct, effects, and the balance of interests. Unfair competition claims also face a high proof threshold. Punitive damages are unavailable.
• Copyright claims are advantageous, offering clear ownership rights, strong exclusivity, a relatively low burden of proof for infringement, and the ability to seek punitive damages; this is because they are enforced as an exclusive right. The disadvantages are raw data and factual information are excluded because protection for that is limited to its expression, and the high threshold for originality. Additionally, copyright protection in China is subject to restrictions on platform rights to user-generated content and time limits on protection.
• Trade secret claims are also exclusive rights, with advantages including broad protectable subject matter encompassing technical information and business data, works and non-works, data sets, and individual data records, along with the ability to seek punitive damages and indefinite protection if secrecy is maintained. The disadvantages are strict secrecy requirements and a high evidentiary burden, requiring proof of secrecy measures, that the information is generally unknown, and that it was acquired by improper means. Relying on this hinders data transferability.

Challenges in Enforcing Data Asset Rights

Collecting evidence of data misappropriation. Data infringement is usually noticed due to suspicious activity, but courts instead look at evidence of unfair use. Under Supreme Court Guiding Case No. 262, even if evidence of data scraping is presented, the plaintiff must also show that the data collection caused significant harm, such as degrading their service,[2] which can come as a surprise for businesses. Moreover, proving unfair data collection or use is challenging and uncertain, often requiring extensive log analysis and tests. Meeting the burden of proof requires a deep understanding of technical logic, evidentiary standards, and types of evidence.

Valuing data assets is also difficult. Courts determine data asset valuation loss as discretionary or statutory damages, and not proportionate damages. The 2025 Unfair Competition Act §22 requires damages to be awarded based on the plaintiff’s actual loss or the defendant’s gain,[1] which can lead to conflicts in the calculation methods.

Here is an overview of what avenues are available for protecting rights in business data:

Strategies

• Layered Protection Tactics
• For critical confidential data, use secrecy measure to support trade secret claims.
• For structured data, consider copyright registration or protect it as a compilation.
• For public or semi-public datasets, look for evidence of unfair competition.

Avoid Inappropriate Tactics

• Do not make trade secret claims for data products publicly sold, as courts generally view such information as non-confidential.
• Do not assert copyright over non-original raw datasets, such as scraped ecommerce price lists or user-generated content.

Key Documentation

• Critical evidence includes nondisclosure agreements, activity logs, risk controls, watermarks, and API request records.
• New Challenges for Data Assets Posed by AI

Unfair Competition Issues Related to Data Use in AI Training

AI training requires enormous amounts of data, and the Chinese legal community is investigating what the rules should be for acquiring and using data. Using copyrighted works, courts in China and elsewhere has typically been held to constitutes fair use. In the Beijing Case 0491-cv-12142 (2023),[5] for example, the court held that using an individual’s voice can be used for model training, with permission.

For data that does not qualify as a copyrighted work, is a license or authorization required for use in AI training? No Chinese court has issued a final judgment, but several closely watched cases are currently pending.

Factual questions are challenging as the legal ones. Chinese courts mainly use evidence of unfair use of data to establish unfairness. AI model training is different because it occurs in secret on computer systems, without outside to observability. So, it’s hard to prove infringement. A question is whether courts will adjust their fact-finding approach as a wave of cases emerges where the only evidence available concerns unfair data acquisition rather than use.

Does an AI Agent’s Use of User or App Data Constitute Unfair Competition?

In China, which has its own tech ecosystem, Qwen Agent has driven three technological approaches to mobile AI agents. The Claude Cowork and OpenClaw desktop agents from early 2026 have exploded in popularity in China, and extended functionality from personal assistance to workplace productivity.

As technology evolves, seamless integration between agents and apps is inevitable. In terms of social stability, however, the potential disruption to the business ecosystem from new digital technologies raise legal questions about whether this new form of competition could be deemed unfair.

There is indeed a possibility that courts may in the future rule that an AI agent’s app use is unfair access to and use of data. From the perspective of Chinese law, an AI agent acts in an agent-principal relationship when it executes on user instructions and represents the user. When it uses an app, it does so by executing a user instruction that requires retrieving data from the app. Digital data is generally controlled and owned by app businesses, while users possess only personal data protection rights under China’s privacy laws, and those are not data asset rights. Chinese app EULAs usually provide that only human users may use the app, prohibiting wrappers, scripts, and automations. All of these factors support unfair competition claims.

A salient question is what rules an AI agent should follow when collecting, accessing, using, and storing data to avoid allegations of unfairness. These issues will be a challenge for international businesses operating in China. What is, or is not, allowed is still unclear since the law has not settled the issue yet; if exposed to significant risks in this area, the strongest mitigation strategy is to find an advisor familiar with enforcement trends and directions.

While it can be hard for international businesses to independently find reliable local attorneys. Even if you find the suitable attorneys, they rarely speak English. Fortunately, CBL has an extensive network of qualified Chinese attorneys and provide them with the legal operations and language translation support necessary for them to succeed in international client matters.

FURTHER READING

Get more insights about protecting data assets in China:

Deter China Employee IP Theft with Referral to PRC Prosecutors

FOOTNOTES

[1] Unfair Competition Act, (中华人民共和国反不正当竞争法), (NPC, Jun.27, 2025), (in Mandarin)

[2] Beijing Case 73-cv-1011 (2021) Supreme Court Guiding Case No. 262, ((2021)京73民终1011号), (Supreme Court, Aug. 28, 2025), (in Mandarin)

[3] Shanghai Case No. 73-cv-490 (2024), ((2024)沪73民终490号), (China Judgements Online, Jul. 30, 2025), (in Mandarin)

[4] Shanghai Case 02-cv-11028 (2023), ((2023)沪02民终11028号), (China Supreme Court, Aug. 28, 2025), (in Mandarin)

[5] Beijing Case 0491-cv-12142 (2023), ((2023)京0491民初12142号), (Beijing Daily, Mar. 21, 2025), (in Mandarin)