On Wednesday, the Trump administration threw its support behind OpenAI, backing the AI giant in its ongoing litigation with The New York Times. The Department of Justice voiced its support through a statement of interest (SOI) filed with the Manhattan judge overseeing the case, which includes a dozen other plaintiffs, including daily papers owned by Alden Global Capital, high-profile book authors, and smaller nonprofit newsrooms like The Intercept.
The filing says the DOJ’s decision to intervene is an attempt to ensure that American AI companies remain “competitive” and are free to “develop a robust AI industry.”
The filing makes grand pronouncements about the harms that could come from a ruling in favor of The New York Times and the other plaintiffs in the news industry. A court finding that training large language models on copyrighted material isn’t covered by “fair use,” the DOJ says, would “threaten national security,” “give a competitive advantage to foreign adversaries,” and thwart “creative and scientific progress while hindering American prosperity.”
“The Administration is siding with a handful of trillion-dollar companies at the expense of the countless American creators whose work they stole,” said Graham James, a spokesperson for the Times, said in a statement. “Both AI and creators can thrive — AI companies simply need to pay fairly for the content that makes their products possible, as copyright law requires.”
“If the administration’s position was accepted, it would result in an unprecedented, uncompensated transfer of IP rights from news organizations to tech companies,” said Matt Topic, a lawyer at Loevy & Loevy who is representing The Intercept in the case.
In its review of case law, the Trump administration largely reproduces a reading of “fair use” that has been put forward in court by many of the biggest AI players, including OpenAI. It argues that LLM development is “transformative,” meaning that the model and its outputs are different enough from original news articles and other content ingested from new publishers to constitute something “new,” or create a different “utility” than the original works.
OpenAI’s models aren’t trying to “directly entertain or educate” the way a Times article does, the administration argues, but are instead performing tasks like writing, translating, and solving problems.
The filing acknowledges the Times’ claim that ChatGPT has been used to reproduce the Times’ copyright material in its entirety. But it dismisses that concern as irrelevant, arguing that the litigation concerns the training of LLMs, not their outputs, and that the models’ most common uses should be separated from “how persons might hypothetically or anomalously” use ChatGPT.
The filing also cites the Times’ own LLM use as a reason to side with OpenAI in the case, arguing that LLMs “help people, including those working in creative fields, make things and get things done.”
I’ve reported for Nieman Lab on how the Times has used LLMs in Pulitzer-winning investigations and to source leads for beat reporters. Sometimes, the newsroom has used GPT models. The filing, however, cites and quotes a Futurism article, arguing that it shows “authors at the New York Times itself are using LLMs to help them ‘conceptualize and edit’ articles.”
That phrase from Futurism’s story actually references a controversy earlier this year when a freelance contributor used AI to structure and edit her first draft of a “Modern Love” column. The Times requires substantial AI use to be disclosed to readers, even from freelancers, but no disclosure appeared in that column and it’s unclear if the freelancer’s editors knew about her AI use before the column was published. It’s not a particularly compelling argument for how the Times is reliant on AI.
The DOJ also argues that a ruling in favor of The New York Times would hurt small and independent news publishers. Underresourced newsrooms can use LLMs “to compete,” the filing says — “by, for example, using an LLM to generate an image to accompany an article — which otherwise might require a photographer or license” or by “directing users to dissenting sources that offer contrary information.” (The Los Angeles Times famously did just that last year when it published an AI-generated counter perspective on the Klu Klux Klan beneath an opinion column.)
Another argument in the filing is that if OpenAI loses, all AI companies would need to pay licensing fees, disproportionately benefiting legacy media outlets due to “the sheer volume of their written publications.”
“It is not in the public’s interest for the largest technology companies to have an oligopoly on LLM training due to licensing entry barriers that function primarily as large subsidies for old mainstream media companies,” the DOJ claims.
It’s an odd position to take. Yes, anyone following the current state of AI licensing deals would tell you that small news publishers have barely seen a dime from OpenAI, even though much of their content has still been scraped to train LLMs. In many other countries, OpenAI has only signed deals with one of the largest news organizations in the country, leaving the rest behind. But legal precedents requiring payment for content could create leverage and new revenue streams for small publishers, since they have rarely been invited to the table to date.
Few small and nonprofit news organizations are suing OpenAI because litigation is so expensive. Many are concerned with copyright violations, but have let the Times take the lead on setting legal precedent as the most resourced newsroom in the U.S. (The Times has spent more than $30 million on AI-related litigation.)
“[The lawsuit] is not simply moral, it’s for the ultimate sustainability of these news organizations and these unique voices,” David Barlow, general counsel for The Intercept, another plaintiff in the Times case, told me in March 2024. “The principle of being able to take content without compensation is critically important to us — and so is our survivability. They go hand in hand.”
