<![CDATA[The rapid development of generative artificial intelligence has increased the use of copyrighted works as training data for AI models, creating legal issues concerning copyright protection and legal certainty for the parties involved. The absence of explicit legal provisions governing the use of copyrighted works as training data under Indonesia's Copyright Law No. 28 of 2014 has intensified the need for a legal liability framework capable of balancing copyright protection with technological innovation. This study aims to analyze legal liability for the use of copyrighted data in artificial intelligence model training under Indonesian copyright law by considering recent international regulatory developments and legal practices. This study employed normative legal research using statutory, conceptual, and comparative approaches through the analysis of legislation, judicial decisions, legal doctrines, and relevant academic literature. The findings indicate that artificial intelligence model training involves the reproduction of copyrighted works that may generate legal consequences, while the current Indonesian legal framework does not provide explicit legal certainty regarding the permissibility of such training activities. The study also finds that legal liability should take into account the respective roles of model developers, dataset providers, distribution platforms, and end users, supported by mechanisms for dataset transparency, text and data mining regulation, and appropriate licensing schemes. This study concludes that reform of Indonesian copyright law is necessary to establish legal certainty while maintaining an appropriate balance between copyright protection and the development of artificial intelligence.]]>
