The Beijing Intellectual Property Court recently issued a landmark ruling that gives new weight to the protection of AI models under China’s Anti-Unfair Competition Law. In this case, Beijing Douyin Technology Co., Ltd., the operator of the popular Douyin app (known internationally as TikTok), sued Yirui Ke Information Technology (Beijing) Co., Ltd., the operator of the B612 KakaoTalk app. The dispute centered around the imitation of a feature developed by Douyin, which transforms photos and videos of real people into a stylized comic look—a function that has become highly popular and has given Douyin a noticeable competitive edge.
Although the court denied a copyright infringement claim, ruling that the comic transformations did not qualify as creative works under the Copyright Law, it unequivocally held that the unauthorized use of the model’s structure and training parameters constituted unfair competition. A key factor was that Douyin had made substantial investments and undertaken considerable development work in creating the model, which forms a core part of the app’s market success. The defendant’s use of Douyin’s specific model architecture was deemed a clear violation of generally accepted principles of business ethics in the field of artificial intelligence, particularly because such structures, refined through extensive training and expertise, are not freely available for use.
The court emphasized that the strong similarity of the effects on both platforms highlighted the substitutability of the products and had caused significant economic harm to the plaintiff. Thus, the court not only confirmed the protection of an innovative technical achievement but also established a new legal precedent: for the first time in China, the structure and parameterization of an AI model were recognized as protectable business interests. As a result, the protection of competitive advantages is now possible even independently of traditional copyright law. For AI developers, this ruling means that future success will depend heavily on thorough documentation of their development processes.
The court also made it clear that mere technical cracking or reverse engineering is not sufficient to invalidate a model’s protection; rather, it is the defendant’s responsibility to prove that the model structure or parameters were lawfully obtained. This decision thus represents not only a milestone in the protection of AI-driven innovations but also a strong signal in favor of fair competition and sustainable innovation in an increasingly technology-driven market environment. It highlights the growing importance of clear legal frameworks and offers companies valuable guidance on how to effectively safeguard their technological developments and investments.
China Intellectual Property Blog 