AI 커버곡의 지식재산권 침해 쟁점과 규율 방안

A Study on Intellectual Property Infringement Risks and Legal Issues of AI-Generated Cover Songs

The rapid development of generative AI is reshaping the production, distribution, and consumption of music. With the widespread use of voice conversion and vocal synthesis technologies, AI cover songs—where users recreate the distinctive voices of real artists—have proliferated as a new form of fandom-driven and meme-based content. Yet this phenomenon raises overlapping legal issues, including potential copyright and neighboring rights infringements, unauthorized use of vocal identity implicating publicity and voice-likeness rights, and possible trademark or unfair competition violations through names, thumbnails, and metadata. Regulatory approaches differ widely across jurisdictions. The United States is moving toward federal publicity-right legislation; the EU imposes transparency and deepfake disclosure duties under the AI Act but lacks clear rules on vocal imitation; Japan separates broad allowances for data mining from personality and unfair competition concerns at the output stage; and China enforces a strong, multi-layered regime directly regulating synthetic voice and image content. Korea, by contrast, remains reliant on partial applications of existing IP and unfair competition laws, without an integrated framework tailored to AI-generated vocal works. This study analyzes the technical structure and usage of AI cover songs, examines the associated legal issues across copyright, neighboring rights, publicity rights, trademark law, and unfair competition law, and identifies regulatory gaps through comparative analysis. Based on these findings, it proposes four directions for reform: a dual framework distinguishing training and output stages; a comprehensive personality-identifier protection regime covering voice and timbre; mandatory labeling of AI-generated content; and platform obligations for monitoring and rights-holder protection. AI cover songs highlight both new creative possibilities and the need to reconsider established norms in music and IP law. The study underscores the importance of developing a coherent and predictable legal framework suited to the generative AI era in Korea.