Healthcare AI 2025

CHINA Trends and Developments Contributed by: Gil Zhang, Diana Li, Muran Sun and Huihui Li, Fangda Partners

ronment underscores China’s broader shift towards a “validation-first” model, where AI healthcare compa - nies are expected to deliver not just visionary technol - ogy, but also demonstrable medical value and regula - tory robustness. Intellectual Property Protection in Healthcare AI Market trends AI is bringing about a profound change in the global healthcare landscape. Innovations such as surgical robots, remote diagnosis, smart diagnostics and wear - able devices are accelerating both in development and deployment. Meanwhile, China has emerged as the global leader in healthcare AI IP rights in recent years, especially in patents. According to the 2025 AI Index Report published by Stanford HAI, in both 2022 and 2023, Chinese entities accounted for more than 60% of global healthcare AI patents, ahead of all other countries. The development of China’s healthcare AI field is driven by the co-operation between enterprises and academic institutions. As mentioned in the IP Press’s White Paper, Ping An Health, for example, is the global leading patent applicant, with 4,176 healthcare AI pat - ents; Tencent comes next, with 1,707 patents nation - wide concentrated in medical imaging and disease detection. Academic institutions are also essential players, exemplified by Tsinghua University’s “Agent Hospital” project deploying 42 AI physicians trained to manage over 300 diseases. Co-developed by Ali - baba’s DAMO Academy and Zhejiang University, the DAMO GRAPE model represents a major milestone as the world’s first gastric cancer screening AI. Unfolding critical issues Healthcare AI innovations in China are eligible for patent protection if they meet the technical criteria set forth by the Patent Law, and applicants should provide detailed descriptions of technical features, highlights of concrete technical effects and substan - tial data support to meet such criteria. However, chal - lenges persist, particularly around the relatively vague examination standards and the conflicts between the transparency of AI models and confidentiality requirements. Meanwhile, the underlying algorithmic logic and structure of training models generally do not meet the threshold for authorship under copyright law

and thus lack direct copyright protection. As a result, healthcare AI companies tend to rely more on trade secret protection to safeguard core models, param - eters and data preprocessing workflows. Concerning copyright protection related to AI-gener - ated content (AIGC), a series of landmark cases have been heard by the Chinese judiciary. All these cases illustrate Chinese courts’ proactive yet inconsistent efforts to adapt traditional IP frameworks to AI gov - ernance. • Regarding whether AI-generated content quali - fies for copyright protection, the Beijing Internet Court, in the nation’s first case involving “AI text- to-image”, recognised the AI-generated image titled “The Spring Breeze Brings Tenderness” as a copyrightable work. • The Suzhou Intermediate People’s Court reached the opposite conclusion in a case concerning a series of AI-generated images of the Phantom Wings Transparent Art Chair, holding that the out - put lacked the requisite originality and personalised expression to qualify as a protected work under copyright law. • With respect to platform liability for infringing AI-generated content, Guangzhou Internet Court, in the world’s first AIGC platform liability case, underscored the reasonable duty of care incum - bent upon AI platforms to monitor and prevent the generation and dissemination of infringing content. Separately, in a May 2025 ruling, the Beijing Haid - ian District Court found that a healthcare company’s unauthorised replication of patient review data and doctor-authored medical educational articles from a competing internet healthcare platform constituted unfair competition, awarding CNY2.3 million in damag - es. The court held that defendant’s conduct, crawling and displaying data from the plaintiff’s platform on its own, violated the principle of good faith and disrupted fair market competition, thereby breaching the general clause of Article 2 of the Anti-Unfair Competition Law. While this case did not directly address AI training sce - narios, its legal principles carry significant implications for the healthcare AI sector. As medical AI development increasingly depends on clinical data resources, courts are placing greater emphasis on protecting platform

CHAMBERS.COM