Trump Administration Sides With OpenAI in New York Times Copyright Lawsuit

Kate Knibbs and Maxwell Zeff report the Trump administration submitted a letter supporting OpenAI against the New York Times' 2023 copyright suit over training on Times articles without permission. The government says the United States has a 'strong interest' in whether AI training is fair use because it connects to retaining global AI leadership. It calls the Times' fair-use definition inconsistent with current law and warns that constraining LLM development would thwart progress and American prosperity. The arguments are meant to apply to related publisher and author cases too.

The letter calls training on copyrighted works 'extraordinarily transformative' and says OpenAI's LLMs do not significantly compete with Times articles. Conflating training with outputs would, it argues, imply it is illegal for humans to train on others' words — compared to a teenage Joan Didion typing Hemingway. IP lawyer Evan Brown told WIRED Judge Sidney H. Stein is not obliged to follow the letter but will take a DOJ filing seriously. Pamela Samuelson called it a significant development consistent with rulings so far. A Times spokesperson said the administration is siding with trillion-dollar AI companies at the expense of creators. Authors Guild CEO Mary Rasenberger called the letter replete with faulty arguments. Context: Meta technically won Kadrey; Anthropic must pay $1.5 billion for pirating books while training itself was held fair use; Sony and Warner sued Anthropic last week.

This is not a ruling. It is the executive branch putting a thumb on fair use as industrial policy. 'Global leadership' is doing the legal work that 'transformative' used to do alone. The Didion/Hemingway analogy is the tell: the government wants judges to treat ingestion as reading, not as copying into a commercial model. Times lawyers will treat it as copying. Stein can ignore the letter and still feel it.

Samuelson noting consistency with existing AI copyright outcomes is the cold read. Alsup already split training (fair) from piracy (billions). Kadrey failed on evidence of harm. The Times case is the high-profile leftover. A DOJ letter does not make OpenAI win. It makes a Times win look like a fight with the administration's AI strategy.

Signal Desk's read: the administration is picking the stack over the stack of newspapers. That is coherent if you think the national interest is model training at scale. It is also a wealth transfer argument dressed as fair use. Rasenberger is not wrong that the letter is a policy document. The $1.5 billion Anthropic check is the reminder that 'training is fair use' has not meant 'anything goes' — shadow libraries still blow up. Sony and Warner arriving last week is the next theater, music instead of newsprint.

Watch whether Stein invites more government paper or treats this as a one-off amicus-without-being-amicus.

Dozens of training-data suits are live. The two landmark 2025 outcomes left a split: training can be fair use, acquiring the books from pirate libraries is not. The Times case is where news publishers try to get a different answer on journalism, not novels.