The U.S. government has weighed in on one of the most closely watched questions in artificial intelligence: whether large language models can be trained on copyrighted material without permission. In a court filing connected to the New York Times' case against OpenAI, the administration argued that the country has a strong interest in building a competitive AI sector.
The brief says maintaining global leadership in AI remains a strategic priority, and that limiting model training too broadly could slow scientific and creative progress. The argument centers on the idea of fair use, a copyright principle that can allow certain uses of protected works when they are considered transformative.
The debate matters far beyond one lawsuit. Systems behind tools such as ChatGPT, Claude, and Gemini are trained on vast collections of text, including books, articles, and other published material. Publishers have increasingly challenged whether that practice should require licensing agreements.
Recent court decisions have already shaped the conversation. In one notable case, a judge approved a major settlement involving Anthropic, while also drawing a distinction between model training and the use of illegally obtained books. That reasoning has become part of the wider legal and technical discussion around how AI learns from data.
The administration's filing does not decide the case, but it adds institutional weight to the broader argument that AI development should remain open, competitive, and innovation-driven. The outcome could help define how future models are trained and how copyright law adapts to the next era of intelligent systems.
