Fair Use in AI Training: What is Considered "Transformative?"

A fascinating legal case which captured the attention of the AI world in 2025, Bartz v. Anthropic sought to clarify the question of whether usage of copyrighted books to train AI models is protected by the “fair use” doctrine. In Episode 12 of The AI Arena, our host Sean interviews Robert Brauneis, Professor of Law and Co-Director of the Intellectual Property Program at George Washington University Law School. Professor Bruneis highlights key facets of Bartz v. Anthropic and Kadrey v. Meta Platforms, Inc., where both judges ruled in favor of AI developers, but with caveats that have since resulted in a potential settlement* and  further questions regarding what is exactly considered “transformative” when it pertains to AI training. Sean skates to the puck alongside Professor Bruneis, who breaks down the four factors of fair use in the US, and offers his opinion on the rulings from both cases. This is a can’t-miss discussion for copyright and patent attorneys working with AI developers, law students interested in this rapidly evolving sector of the profession, and organizations seeking clarity on the usage of the technology in their day-to-day operations.  *Note: This episode was recorded before public disclosure of a potential $1.5 billion settlement by Anthropic to book authors.