Generative AI and Copyright Law: The De Facto End of the Berne Convention Era and The Need for a Shift in the Public Paradigm

The advent of generative AI is bringing fundamental changes to the exclusivity of copyright and the principle of territoriality, thereby requiring a shift in the public paradigm. Measures currently under discussion, such as text and data mining (TDM) exceptions or fair use, can hardly be regarded as appropriate frameworks to secure copyright protection at the training stage. Some have argued that the matter should be left to contractual arrangements between AI companies and copyright holders, but such an approach is insufficient as a policy response for public authorities charged with cultural governance. To make copyright protection effective during the training phase of generative AI, a substantial paradigm shift in the understanding of copyright’s exclusivity is necessary, and given the transnational character of AI services, exceptional approaches to the principle of territoriality are required to ensure fair competition for domestic enterprises. In the short term, the most effective response may be to guarantee proportional remuneration for copyright holders through the expansion of statutory licensing schemes, or an equivalent presumption of licensed use, at the training stage. In the longer term, improvements to the collective management system will be needed to reconcile copyright protection with technological progress, while at the international level, active participation in the establishment of a new regulatory framework to replace the Berne Convention will be essential. Copyright law has, from its inception, functioned not merely as a regulation of private rights but as a cornerstone of cultural policy and a public regulatory framework. In the era of generative AI, its leading role is more important than ever for ensuring the sustainable development of culture and cultural industries.