US Government Sides With OpenAI in NYT Case — Argues Training Is 'Transformative Fair Use'

The executive branch has picked its side in AI's defining copyright fight. The Trump administration contributed a 20-page statement of interest in The New York Times' lawsuit against OpenAI, arguing that training LLMs on copyrighted text is transformative fair use (TechCrunch). It is not a ruling, and its authors hold no jurisdiction over the Southern District of New York — but it is the government's most explicit endorsement yet of the doctrine OpenAI's training pipeline depends on. Our read: no operational change today, and the SDNY ruling — not this brief — remains the decisive event. The filing's real weight is political: it reframes the case from "did OpenAI steal?" toward "would a ruling against OpenAI throttle American AI?"

What happened

TechCrunch reported on September 2 that the administration filed the brief in the SDNY case, backing OpenAI against NYT's claims that the company trained on the paper's articles without permission. The filing argues the United States has "a strong interest in continuing to develop a robust and competitive artificial intelligence industry that sets the standard for the practice and procedure of AI use globally," and leans on the executive order Trump signed last year committing the US to "retain global leadership in artificial intelligence" (TechCrunch). Its core claim:

"Constraining LLM development under a misunderstanding of fair use doctrine would thwart such creative and scientific progress while hindering American prosperity and economic mobility." — US statement of interest (via TechCrunch)

The brief lands in a litigation environment that has been running favorable to AI companies on the fair-use question. Last year Judge William Alsup approved a $1.5 billion settlement between Anthropic and a group of writers — a penalty aimed at Anthropic's use of illegal shadow libraries to pirate training books, not at the act of training itself (TechCrunch). Alsup likened an LLM's training to a human reading a book: "Like any reader aspiring to be a writer, Anthropic's LLMs trained upon works not to race ahead and replicate or supplant them — but to turn a hard corner and create something different" (TechCrunch).

The limits matter as much as the endorsement: the brief is not binding, the case is being tried in the Southern District of New York, and the executive branch does not control that courtroom (TechCrunch).

Why it matters

For every AI builder the training-data question is existential: if unlicensed training is not fair use, the entire frontier model stack is retroactively illegal — and so is every product built on it. This filing is a signal as much as a legal move. The executive branch is telling the courts and the publishing industry that it views LLM training as the kind of transformative use copyright law exists to protect. Non-binding filings still move judges, and this one hands the NYT court a geopolitical frame for a doctrine question.

What changes for you

• Nothing operational today. No verdict change, no code change, no compliance deadline — the statement of interest is advocacy, not law.
• If you build on frontier APIs: the filing trims — it does not eliminate — the tail risk that training-data liability cascades into the models you depend on. The decisive events remain the SDNY ruling and whatever follows on appeal.
• Publishers and content-licensing businesses: expect the government's fair-use position to harden the other side's negotiating stance at the licensing table.