The Justice Department filed an amicus brief on September 1 in the U.S. District Court for the Southern District of New York supporting OpenAI’s view that training large language models on copyrighted text can qualify as fair use, marking what is widely seen as the U.S. government’s first formal stance in AI-training copyright litigation. The filing addresses multiple suits, including The New York Times’ December 2023 case against OpenAI and Microsoft, and argues that reproducing text for training is highly transformative and should be judged separately from any later outputs that might copy articles. Officials said the United States has a strong interest in courts rejecting claims that such training violates copyright, citing scientific progress and national security, and warned that rules making a robust U.S. AI industry substantially harder to build would threaten security and advantage adversarial nations. The brief also opposes broad licensing-fee mandates that could concentrate development among large tech firms and effectively subsidize major media companies, while stressing that the government is not endorsing any specific conduct in the pending cases. A Times spokesperson condemned the Trump administration for siding with major AI companies over creators and said firms should pay fair compensation; OpenAI did not comment. The brief is not binding, and the court will decide the fair-use questions as dozens of related cases proceed.