The First Case involving Unauthorised Use of Registered Data in China

Yinmu (Shanghai) Technology Co., Ltd. vs. Datatang (Beijing) Technology Co., Ltd. (2024) Jing 73 Civil Final 546

Background

The respondent is a company engaged in data processing, artificial intelligence system services, internet information services, technology development, and other related services. On September 15, 2021, the respondent released the "AI Data Open Source Project: 1505 Hours of Mandarin Chinese Voice Data," offering open-source data to non-commercial organizations such as universities and academic institutions. It explicitly stated, "any form of commercial use without permission is prohibited," and for any compliant usage, users must acknowledge the use of "Datatang AI Dataset" in their research successes and also cite the source. The respondent registered the intellectual property rights for the released data and obtained the registration certificate (No. 2023000007) from the Beijing Intellectual Property Office. The appellant illegally obtained a 200-hour data set from the respondent and provided it for download on its official website (www.graviti.cn), enticing users to register as members.

Three Unsupported Claims of the Respondent

The appellant's actions did not infringe upon the respondent's data property rights, copyright, and trade secrets.

Reading into the Judgment (Analysis)

(a) Data Property Rights: Article 127 of the PRC Civil Code states that data is protected by law, but it does not classify data as property rights under legal protection. The Civil Code adheres to the basic principle of the statutory nature of property rights, and since the Civil Code does not explicitly list data as property rights, it should not be broadly interpreted to grant data property rights protection. The appellate court believes that, according to Article 127 of the Civil Code, the data in question can receive better protection under the Anti-Unfair Competition Law.

(b) Copyright: Whether the respondent's copyright has been infringed primarily focuses on whether the data set in question constitutes a protected compilation work under copyright law. If it does, according to Article 15 of the Copyright Law, the copyright of compilation works is held by the compiler. Here, the appellate court supports the first-instance court's view, stating that the data set in question was compiled from basic voice files (WAV audio files, metadata files, TXT text files), but the structure, distribution, and overall layout within each folder remain the same. The data set does not exhibit unique creativity in its presentation and layout, thus lacking originality in its content selection and arrangement, and does not qualify as a compilation work protected by copyright law.

(c) Trade Secrets: Article 9, Clause 4 of the Anti-Unfair Competition Law defines trade secrets as technical information, operational information, and other business information that is not publicly known, has commercial value, and is subject to corresponding confidentiality measures by the rights holder. Here, "not publicly known" means that the information is not generally known or easily accessible to those in the relevant field. The determination is based on the relevant field's personnel, and the standard is that the information must not be "generally known" or "easily obtainable." The point of determination is when the alleged infringement occurs. Therefore, if information can be easily obtained from public sources, it loses its qualification for protection as a trade secret. In other words, if the holder of the trade secret publicly discloses the information in online or other public spaces, as long as personnel in the relevant field can directly access it, the information loses its secrecy. Thus, the court does not consider the respondent's data set to constitute a trade secret.

One Supported Claim of the Respondent

The appellant’s actions violated Article 2 of Anti-Unfair Competition Law.

Reading into the Judgment (Analysis)

(d) Violation of the Anti-Unfair Competition Law: The appellate court believes that the data set in question is a result of significant investment of human, material, and technological resources by Datatang Company in collecting and organizing, containing a wealth of voice data entries suitable for training artificial intelligence models. This meets the needs of entities developing AI models for voice data, providing them with trading opportunities and competitive advantages. Based on the supplementary principle of Article 2 of the Anti-Unfair Competition Law, regarding the protection of data sets, if they are in the public domain and their content selection and arrangement contribute originality, they should be primarily protected as compilation works. Conversely, if the data set is not easily accessible to personnel in the relevant field, trade secret protection may apply. If the data set is in the public domain and lacks originality in its content selection or arrangement, due to the absence of exclusive intellectual property rights and the foundation for trade secret protection, it may be regulated under Article 2 of the Anti-Unfair Competition Law. The appellate court finds that the appellant has violated the Anti-Unfair Competition Law.

Conclusion

Although this case did not recognize the data asset registration certificate as property rights under the Civil Code, it affirmed the legal protection that the data should receive. Ultimately, the respondent was granted some degree of acknowledgment and protection for violations of the Anti-Unfair Competition Law. As local pilot projects for data asset registration and filing continue to unfold across China, it is possible that legislation may soon be considered to affirm the status of data as property rights, providing clearer and more robust legal grounds for judicial decisions.