<![CDATA[The rapid diffusion of generative artificial intelligence (AI) systems capable of producing text, images, music, source code, and other expressive or functional outputs has exposed deep structural tensions within contemporary intellectual property (IP) law. Doctrines built around human authorship, inventive step, and identifiable infringing acts strain when confronted with machine-generated works, opaque training processes, and diffuse chains of causation between input data and output content. This article examines the legal challenges generative AI poses to copyright, patent, and related IP regimes, focusing on three interlocking problems: the (non-)recognition of AI systems and AI-assisted human creators as authors or inventors; the legality of using copyrighted works as training data without authorization or remuneration; and the adequacy of existing liability and governance frameworks to address infringement risks generated by foundation models. Using a normative juridical method combined with comparative and conceptual approaches, this article analyzes statutory provisions, judicial decisions such as, and recent international scholarship to map divergent regulatory responses across the United States, the European Union, China, India, and Indonesia. The findings indicate that no jurisdiction has yet produced a fully coherent doctrinal settlement; instead, a patchwork of judicial improvisation, administrative guidance, and emerging legislation such as the has developed. The article concludes that IP law requires targeted reform rather than wholesale replacement, including clarified human-authorship thresholds, statutory text-and-data-mining exceptions coupled with opt-out and transparency mechanisms, and graduated liability rules that distinguish developers, deployers, and end users of generative AI systems.]]>
