<![CDATA[The development of Artificial Intelligence in the music industry raises new legal challenges within copyright law, particulary regarding the use of copyrighted musical works as training data for AI algorithms, AI training processes typically involve digital reproduction, dataset storage, and musical pattern extraction, often conducted without consent or attribution, thereby posing risks to the protection of authors moral rights. This article examines whether AI training activities ca be legally classified as reproduction under Indonesia’s Copyright Law (Law No. 28 of 2014) and analyzes their implications for the protection of moral rights. Using a normative legal research method with statutory, conceptual, and limited comparative approaches, particularly referencing the European Union’s regulation of Text and Data Mining and AI governance. This study finds that AI training processes normatively fulfill the elements of reproduction, both temporary and permanent. Furthermore, such practices result in systemic neglect of author’s moral rights duet o the loss of attribution and the opacity of algorithmic system. The article argues that Indonesia’s copyright framework including the 2025 Draft Copyright Bill, remains largely output-oriented and does not adequately regulate the use of copyright works as input for AI training. Therefore, a legal reform is necessary to explicitly recognize AI training as a form of copyright relevant use to ensure effective protection of authors moral rights in the artificial intelligence era.]]>
