Research on the Copyrightability of AI-Generated Works

With the continuous iteration of AI large models like DeepSeek, AI has gradually evolved from an auxiliary tool to a quasi-autonomous content-producing "entity". The massive amount of AI-generated content has triggered a series of legal disputes. Among them, issues whether AI generated output qualifies as a copyright-protected work and the determination of infringement have become increasingly prominent. However, the legal vacuum has not only damaged the rights and interests of creators but also disrupted the innovation ecosystem. Based on this, this paper reviews existing theoretical and practical viewpoints, analyzes the feasibility of copyright for AI-generated works from the perspectives of the existing affirmative, negative, and compromise theories, and explores the theoretical rationale for the copyrightability of AI-generated works from the aspects of the originality of works and meeting the requirements of certain forms of expression, so as to provide a theoretical reference for the possibility of legal regulation of new objects in the era of digital economy in China.