US government backs OpenAI, says AI training on copyrighted text is fair use

US government sides with OpenAI

The Trump administration has formally intervened in The New York Times' copyright lawsuit against OpenAI, filing a statement of interest that supports the AI lab's central defence: that training large language models on copyrighted text counts as fair use. According to The Verge, the filing was submitted this week in a case that could shape how courts treat AI training data across the industry.

The Times sued OpenAI and Microsoft in December 2023, alleging that OpenAI built its systems on the newspaper's journalism without permission and seeking damages it has described as running into the billions. At stake is a question still being worked out in courtrooms: whether ingesting copyrighted material to train a model is a transformative fair use or simply copying at scale.

What the government's filing argues

Government attorneys argued that the Times is effectively asking the court to carve AI training out of fair-use doctrine altogether, an outcome they described as inconsistent with basic principles of copyright law and at odds with copyright's constitutional purpose of promoting progress in science and the useful arts. They also made an economic and scientific case, asserting that large language models are already helping researchers achieve major breakthroughs and that restricting their development through a misreading of fair use would set back scientific progress while harming American prosperity and economic mobility.

Notably, the attorneys conceded that fair-use analysis depends on the specific facts and uses in each individual case. But they contended that imposing broad copyright liability that would make model training impermissible without licences would be legally incorrect and, in practice, unworkable — and that training large language models fits with the creative progress copyright was designed to foster.

A pattern of policy through litigation

The Verge reports that this is not an isolated move. The administration has leaned heavily on statements of interest in private lawsuits as a policy tool, with one official describing the approach as remarkably effective at advancing its agenda. It has previously argued that AI training should count as fair use, including in its National AI Legislative Framework.

There is a personal layer as well. Trump is pursuing an active defamation suit against the Times, and his long-standing hostility toward the paper colours the intervention, as The Verge points out.

A fragmented legal landscape

The filing lands in an already unsettled environment. Other publishers, including the Center for Investigative Reporting, the Chicago Tribune and the New York Daily News, have filed their own lawsuits over AI training data. In a milestone 2025 ruling, a judge found that Anthropic could lawfully train its models on books it had legitimately acquired, while still leaving the company exposed to liability for piracy — a split outcome that ended in a $1.5 billion settlement with authors.

Much of the industry, meanwhile, has chosen licensing over litigation. Dozens of outlets, among them The Associated Press, Axel Springer and Vox Media, have signed deals with OpenAI. The Times itself, even while suing OpenAI, licensed its editorial content — news articles and recipes included — to Amazon in 2025 for use in Amazon's generative AI tools.

Why it matters

A statement of interest does not decide the case, but it signals where the executive branch wants copyright law to land, and it hands OpenAI the weight of the federal government's argument at a pivotal moment. As The Verge notes, the outcome could set precedent for the many other publishers frustrated with AI systems trained on their work. If training on copyrighted text without a licence qualifies as fair use, publishers lose their strongest form of leverage over AI companies; if it does not, AI labs face a wave of licensing demands and damages claims. Either way, the ruling will help determine who gets paid for the data that powers AI systems — and how much control publishers retain over the work they produce.