The DOJ sided with OpenAI against the New York Times (NYT). What that means for Reddit (RDDT) and News Corp (NWSA)
Key points
- DOJ filed in support of OpenAI on Wednesday
- Government calls AI training "extraordinarily" transformative
- Reddit, News Corp and the Times sell training licenses
- Lutnick urged G20 to embrace "fair use" for AI training
The Justice Department has taken a side in the most closely watched copyright case in AI, and it sided with the AI companies. In a statement of interest filed Wednesday in Manhattan federal court, the government told the judge in The New York Times (NYT) v. OpenAI and Microsoft (MSFT) that "the United States has a strong interest in this court rejecting any argument that training LLMs on copyrighted texts violates copyright law," Reuters reported. The brief called training "extraordinarily" transformative and said limiting it would "thwart such creative and scientific progress while hindering American prosperity."
Spokespeople for the White House, the Times and OpenAI didn't respond to Reuters' requests for comment on Wednesday.
Advertisement
Which companies have money riding on the outcome
The Times sued in December 2023 over millions of articles it says were used without permission. It argues that ChatGPT competes with its journalism and costs it subscription, licensing and ad revenue. The licensing part is where public companies are exposed. AI developers have been paying for content in part because the fair-use question was unsettled, and several of those deals are large. News Corp (NWSA) signed an agreement with OpenAI in May 2024 that Variety put at more than $250 million over five years. Reddit (RDDT) receives about $60 million a year from Google, per CNBC, and its shares fell on July 22 after a report that Google might not renew. The Times itself, while suing OpenAI, signed a deal with Amazon that GeekWire reported at $20 million to $25 million a year.
A fair-use ruling on training wouldn't cancel those contracts. The deals also cover displaying content in chatbot answers with attribution and access to live data, which fair use on training doesn't reach. It would change the negotiating position, though. A buyer that can legally train on public material without a license is paying for access and freshness rather than for permission, and that's a smaller product.
What the brief does and doesn't argue
The government's argument is limited to the training phase. It doesn't address the Times' other claim, which is that ChatGPT reproduced near-verbatim copies of its articles. The regurgitation claim is a separate question, and it's the one where the newspaper has filed its most detailed exhibits.
There's precedent for the government's position. Judge William Alsup ruled in June 2025 in the Anthropic authors case that using legally bought books to train Claude was "exceedingly transformative," then denied fair use for pirated copies, and Anthropic agreed to pay $1.5 billion to settle. That settlement received final court approval on July 20, at about $3,000 a book, and it sets no binding precedent. The DOJ brief asks the Times' judge to adopt the training half of that outcome.
OpenAI's public position hasn't changed since the suit was filed: "Training AI models using publicly available Internet materials is fair use, as supported by long-standing and widely accepted precedents."
Advertisement
The G20 message in Chapel Hill
The filing landed the same day Commerce Secretary Howard Lutnick told G20 innovation ministers meeting in Chapel Hill, North Carolina, that they should allow AI companies to train on creators' work while finding a way to "protect artists," Reuters reported. Lutnick said countries should embrace "fair use." "We need frameworks that protect creators and inventors whose ingenuity drives our economies without creating the barriers that prevent the next generation of innovators," he said, without offering specifics on how to do both. The administration is also asking attendees to sign on to what it calls the Carolina Principles, under which countries agree to "reserve new regulation for novel considerations," per Quartz. Nvidia (NVDA) CEO Jensen Huang made a similar point at the same meeting. "Don't regulate hypothetical theoretical harm, regulate actual and pragmatic harm," he said, according to Reuters.
Taken together, the filing and the G20 agenda describe a consistent US position: the models are treated as strategic assets, training data as a cost of doing business, and rights holders as parties who can negotiate for access rather than block development.
Where the stocks are
| Company | Price (12:30 p.m. ET Wed.) | Day change | AI licensing exposure |
|---|
| Reddit (RDDT) | $154.51 | +4.5% | Google ~$60M/yr, OpenAI deal |
| News Corp (NWSA) | $30.73 | +0.1% | OpenAI, $250M+ over 5 years |
| New York Times (NYT) | $68.08 | +1.2% | Amazon $20-25M/yr; plaintiff vs OpenAI |
| Microsoft (MSFT) | $493.87 | -2.6% | Co-defendant with OpenAI |
| Nvidia (NVDA) | $226.92 | +2.8% | Sells the training hardware |
None of the five showed an obvious reaction to the filing. Reddit rose on a mixed day and Microsoft declined on unrelated news. The market is likely to price the question when a judge rules rather than when the government files a brief.
Where I stand
I've been following this case since the Times filed it in December 2023, and I don't have a strong view on how it should come out. It's one of the more intriguing lawsuits in the market right now, because the answer sets the price of training data for everyone, and I'm mostly interested in seeing which direction the court takes it.
We covered the other side of this market yesterday, when Google approached Disney and the studios about paying for characters, and last week when Sony and Warner sued Anthropic's founders personally. Music and film are where the licensing money still lives. After this week, the value of a text license looks considerably lower.
Advertisement
