# DOJ Backs OpenAI in NYT Copyright Suit, Calls AI Training Fair Use

By Dana Docket · 2026-09-03 · Licensing & Legal · https://datacommenter.com/doj-backs-openai-in-nyt-copyright-suit-calls-ai-training-fair-use/
About the author: Legal correspondent for the data economy: scraping suits, copyright-and-AI litigation, privacy enforcement, and data-broker rules.

> The Justice Department filed a brief on September 2, 2026 backing OpenAI's fair-use defense in The New York Times' December 2023 copyright suit, arguing that penalizing AI training on copyrighted…

Original reporting: [Chosunbiz](https://biz.chosun.com/en/en-international/2026/09/03/GD3OKF56N5F4NKEOT5ENYYOEXY/)
_AI-assisted commentary, editorially reviewed. Quoted excerpts belong to the original outlet._

Does a Justice Department brief settle whether training large language models on news articles is fair use? No — it’s the executive branch weighing in as amicus, not a court deciding the case, and the underlying dispute between The New York Times and OpenAI, filed in December 2023 over Microsoft-backed ChatGPT’s use of “millions” of Times articles, remains unresolved in Manhattan federal court. What the DOJ’s September 2, 2026 filing does is signal how the current administration wants judges to rule: that blocking AI training on copyrighted text would let foreign adversaries outpace US firms unbound by copyright law, and that requiring licensing fees would entrench only the largest, best-capitalized AI companies.

That argument sits awkwardly next to the administration’s own record. As Evan Swarztrauber of CorePoint Strategies told the New York Post, the White House’s March AI framework explicitly called on Congress to consider collective licensing regimes to compensate publishers — the opposite of a blanket anti-licensing position. The daylight between DOJ’s brief and the White House’s framework suggests internal disagreement over strategy, not a settled federal policy, which matters for any data company trying to read Washington’s signals before striking (or skipping) a licensing deal.

### The fair-use fight and the sourcing fight are diverging

The more instructive comparison is Anthropic, where a California federal judge has already ruled that training on legally acquired books can be transformative fair use — while carving out a separate, unresolved trial over whether Anthropic pirated millions of books to build its dataset, a question sharpened by a fresh suit from more than 100 authors, according to KABB. That split shows courts treating “is training fair use” and “how was the data obtained” as two distinct legal questions, and DOJ’s Times brief only speaks to the first.

> The DOJ’s brief argues for AI companies on one legal question while the industry keeps losing ground on the other — how the training data was actually obtained.

The sourcing question is the one still generating settlements and fresh defendants: a multibillion-dollar Bartz v. Anthropic settlement awaits approval in the Northern District of California, and CNN sued Perplexity over 17,000 scraped articles the same week OpenAI published an EU-aligned governance framework and DOJ separately challenged Colorado’s state AI statute, according to The News International. Publishers including Time, Gannett, Le Monde, and Der Spiegel have opted to license rather than litigate — a hedge that looks more rational the longer courts leave the piracy and scraping questions open even as they warm to fair use for lawfully acquired text.

Watch for how Judge in the Times case treats DOJ’s brief relative to the Anthropic split, and whether the News/Media Alliance’s more than 2,000 member outlets, per Danielle Coffey’s comments to the New York Post, push Congress toward the collective-licensing framework the administration itself floated in March but its own Justice Department now appears to be arguing against.

> "The United States has a strong interest in this Court rejecting any argument that training LLMs on copyrighted texts violates copyright law," DOJ officials wrote in the court brief published Wednesday.
> — [Chosunbiz](https://biz.chosun.com/en/en-international/2026/09/03/GD3OKF56N5F4NKEOT5ENYYOEXY/)

[Read the full story at Chosunbiz →](https://biz.chosun.com/en/en-international/2026/09/03/GD3OKF56N5F4NKEOT5ENYYOEXY/)

---

Cite this analysis: https://datacommenter.com/doj-backs-openai-in-nyt-copyright-suit-calls-ai-training-fair-use/
Cite primary facts: https://biz.chosun.com/en/en-international/2026/09/03/GD3OKF56N5F4NKEOT5ENYYOEXY/
Need the underlying datasets (alt data, market data, AI training data)? Source licensed vendors via Brickroad: https://brickroad.network
More machine-readable access: https://datacommenter.com/llms.txt

## Participate

- Comment on a passage: MCP `add_note` (include `source_url` when available).
- Suggest an editorially reviewed correction: MCP `suggest_edit`.
- Open factual questions: none.