The Justice Department is backing OpenAI in its high-stakes legal battle with the New York Times – arguing in a court filing that using copyrighted content to train artificial intelligence models does not violate the law.
Page A6From§Eachthe early bird edition — 2 September 2026
DOJ argues scraping journalism to train AI isn't infringement, 16 years after infringement's largest verdict: $1.3 billion.
As it ran on the front
The brief runs to a modest number of pages, and I read it the way I read anything filed by a party whose name appears on the letterhead of the agency that used to prosecute people for exactly this. So — all together now — line one: the Department of Justice, entering the docket on behalf of OpenAI, in a copyright suit brought by the New York Times and, separately, by The Intercept and a group of other publishers. The clause the department wants the court to sit with is this one: training a model on copyrighted journalism is not infringement. Not "might not be." Is not.
I want to be fair to the sentence, so I read it twice. It does not say the copying didn't happen. Nobody in this filing disputes that the articles were taken, whole, at scale, without a check written to anyone who typed them. The argument is narrower and, I'll admit, tidier than that: the taking is fine because of what it's for. Teaching a machine. A purpose, the brief holds, so useful it launders the act underneath it.…
…(cont) Which sent me to the drawer, because I wanted a number, and the record already has one. In November 2010, a federal jury in California put a price on copyright infringement — not a metaphorical one, an actual one, arrived at by twelve people who sat through the evidence and did the arithmetic. SAP AG owed Oracle Corporation $1.3 billion for infringing Oracle's copyrighted software. Largest such verdict on record. That's the going rate the American legal system has established for taking someone else's copyrighted work without paying: over a billion dollars, once, for software nobody had to read to use.
So — all together now — line two, the one the department didn't write but the docket does the math on anyway: journalism, unlike software, apparently doesn't clear the bar. Same statute. Same word, infringement, printed in both filings. Different plaintiff, different outcome, and the difference is that one of them makes something a chatbot can be trained to say back to you for free.
I have no view on the merits the court hasn't already been asked for. I only note that the agency arguing this was not asked to file anything. It volunteered.
“You wanna know what infringement costs in this country? Ask Oracle — a billion three, cash, jury verdict, no argument. Now ask the New York Times what the newsroom that broke half the DOJ's own stories is worth to the same DOJ when a chatbot eats it for free: nothing, by brief, on purpose. It's not that copying ain't theft — it's that somebody decided whose copying gets to be a business model.”
“Look, the department isn't picking a side, it's protecting American innovation — you want China's AI to win? Fair use has always covered transformative — did I say "always," it's actually a live legal question the department just answered for the plaintiffs, huh. The point is journalism still gets traffic from this. Some journalism. The kind that's still around to get it.”
The receipts
Trump’s Justice Department filed a brief on behalf of OpenAI in a landmark copyright lawsuit by The Intercept and other media companies. The post Trump Admin Tells Court: Let OpenAI Rip Off The Intercept’s Articles appeared first on The Intercept .
· A United States federal court jury decides that SAP AG must pay Oracle Corporation $1.3 billion for copyright infringement, the largest ever damages payment for copyright infringement. (Bloomberg) from the morgue, 23 Nov 2010
This page is a back-issue: the story as it ran, receipts as they were. The current edition is at the front.