Overview
- The Department of Justice filed a 20‑page statement of interest in the Southern District of New York arguing that copying copyrighted text to train large language models can qualify as fair use and is “exceedingly transformative.”
- The filing, which multiple outlets reported was submitted on Tuesday, separates the legal question of training from the question of whether particular AI outputs infringe copyrighted works and urges the court not to collapse the two issues together.
- The New York Times condemned the brief, saying the administration is siding with a handful of trillion‑dollar AI companies instead of creators and urging that AI firms should pay fairly for content used to build their products.
- The DOJ framed its position as a policy choice for U.S. competitiveness and national security, warning that mandatory licensing or broad liability for training could raise entry barriers and advantage foreign competitors or only the largest tech firms.
- The filing is nonbinding and comes into a fragmented legal landscape where recent rulings have diverged—most notably a judge who called training transformative while authors won a roughly $1.5 billion settlement with Anthropic—so the court’s decision will be pivotal for future licensing and industry practice.