Abstract:The widespread adoption of generative artificial intelligence technologies has spurred a wave of secondary creation, typified by "AI-altered" videos . While these creations circulate extensively, they also pose significant copyright risks and governance challenges. Following the creative logic of "AIaltered" videos, this paper conducts a normative analysis to explore the legal characterization of prompts prior to creation and the compliance challenges associated with copyrighted training data. It systematically examines the legal attributes of such videos during the creation process and the risks of infringing on multiple copyrights, including the rights of adaptation and reproduction, while rigorously scrutinizing the "fair use" defense. The paper further argues that platforms should bear enhanced duties of care in the post-creation phase, shifting from passive responses to proactive prevention. To effectively address secondary creation controversies such as "AI-altered" content, which are prone to copyright disputes, it is essential to promote multidimensional coordination, including legal refinement, the establishment of industry standards, innovation in licensing mechanisms, and public education. A systematic governance framework integrating law, technology, and collaborative governance should be developed to achieve a sustainable balance among copyright protection, technological innovation, and cultural preservation.