The Copyright Challenges in the Era of Large Language Models

On May 12, 2026, Maja Bogataj Jančič was a keynote speaker at the Joint Workshop on Legal and Ethical Issues in Human Language Technologies (LEGAL2026) and Computational Approaches to Language Data Pseudonymization, Anonymization, De-identification, and Data Privacy (CALD-pseudo 2026) in Palma de Mallorca.

Large language models (LLMs) learn by consuming culture on a large scale – books, articles, code, and the accumulated product of human creativity. Copyright law is therefore increasingly portrayed as the enemy of AI progress. What if copyright is not an obstacle but a necessary legal infrastructure?

Maja presented the global copyright framework for building generative AI, focusing specifically on Europe’s exceptions for text and data mining (TDM). As courts, regulators, publishers, and AI companies grapple with the issue of training data, the real question is not whether AI will use copyrighted works, but who will control the knowledge infrastructure on which generative AI depends. Europe’s approach is unique and important because the scientific research exception for TDM provides legal ground for building Public Interest AI.

You are invited to view the entire presentation here.