Originally published June 2026. Edited on Jun 27, 2026.
Two and a half years into the lawsuit, the Times's archive remains largely outside ChatGPT and the other major answer engines. Whether that's a costly mistake or a long-game move depends on what the SDNY ultimately rules — and when.
The New York Times appears to have accepted reduced visibility inside some AI systems as the cost of pursuing a broader legal challenge. It is a defensible bet — and arguably the right one for a subscription business that owns one of the most valuable archives in the English language. But the cost is already being paid, in real time, regardless of how the case ultimately resolves.
The Times sued OpenAI and Microsoft in December 2023. Two and a half years later, the lawsuit is grinding, the precedent is unsettled, and the Times has not yet licensed a single major AI engine where consumers now look for answers.
ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews generally surface Times content less frequently than the publication's historical authority might suggest — with one exception: a deal with Amazon, signed in May 2025, that does not put the Times back inside ChatGPT.
NYT v. OpenAI — Litigation Timeline
| Date | Event | Source |
|---|
| Dec 27, 2023 | NYT files complaint in SDNY | CourtListener docket |
| 2024 (mid) | Court denies major dismissal motions; case proceeds to discovery | SDNY docket |
| Oct 2024 | NYT issues cease-and-desist to Perplexity | NYT |
| May 29, 2025 | NYT–Amazon licensing deal announced ($20–25M/yr est.) | WSJ · Amazon |
| Jan 5, 2026 | Judge Sidney Stein affirms order compelling 20M-log sample production | SDNY docket |
| Mid-2025 | Bartz v. Anthropic and Kadrey v. Meta rule AI training can be fair use | N.D. Cal. dockets |
Why the Times sued
The legal theory is real. So is the business reason. OpenAI trained ChatGPT on copyrighted Times content without permission and without a license. The Times's complaint alleged the model had memorized substantial portions of Times articles and could regurgitate them on prompt. That regurgitation evidence is the heart of the case.
Why the lawsuit is grinding
Federal litigation against well-funded defendants on novel legal questions does not move fast. Two and a half years in, the case has cleared the motion-to-dismiss phase and is deep in discovery. In 2025, the Times demanded 1.4 billion private ChatGPT user conversations. The demand was scoped down to a 20-million-conversation sample. On January 5, 2026, Judge Sidney Stein affirmed the order compelling OpenAI to produce the full 20-million-log sample.
In two separate federal cases in mid-2025 — Bartz v. Anthropic and Kadrey v. Meta — courts ruled that AI training on copyrighted material can be highly transformative and protected by fair use. Neither ruling is binding on the Times case, but they cut against the Times. Full landscape: Lawsuit War Room.
What's on the table, what isn't
In May 2025, the Times signed its first major AI licensing deal — with Amazon. Reported by the Wall Street Journal to be worth roughly $20–25M/year, covering NYT, NYT Cooking, and the Athletic — for Alexa, Rufus, and Amazon's foundation models. A smaller Apple agreement was announced earlier in 2025.
The Times is not opposed to licensing in principle. The Times is opposed to licensing OpenAI specifically — the company it sued, the company whose core product is the consumer default for AI-assisted research, the company whose deal terms would have set the floor for every subsequent publisher negotiation.
Why News Corp chose the opposite strategy
News Corp made the opposite calculation in May 2024. The reported $250+ million OpenAI deal covered the Wall Street Journal, the New York Post, Barron's, MarketWatch, the Times of London, and dozens of other titles — and it bought News Corp three things the Times's lawsuit cannot: cash, distribution, optionality.
The compounding cost
By multiple recent measures, more than a third of consumers now begin product research with an AI engine rather than with Google. When the Times is less visible in those answers, the gap goes to whichever outlet covered the same story and is retrievable — in practice, often the News Corp portfolio, the Associated Press wire, the Financial Times, the Atlantic, Reuters, Vox, the Condé Nast portfolio. Tracked in the AI Licensing Tracker.
What if the Times is right?
Suppose the Times wins. A ruling against OpenAI on the core question forces every major AI company to either license at scale or pull training corpora and rebuild. Existing deals get repriced. Statutory damages under the Copyright Act run up to $150,000 per willfully infringed work. The Times's archive is enormous. The publishers who took the deal early did so on the assumption the legal question would resolve favorably for AI companies. If the Times wins, that assumption collapses, and the publishers who took the deal get a second negotiation on stronger ground.
What a Times loss would do
If the regurgitation evidence is not enough to distinguish from Bartz and Kadrey, the most likely outcome is settle-and-license — but two, three, or four years after the publishers who signed early. By that point, consumer citation patterns will have hardened.
Why this matters for brands
The New York Times still matters. The Times still moves narrative inside print, inside legacy broadcast, inside the social distribution layer, and inside the subset of AI surfaces — Alexa, the Times's own search products, Apple's surfaces — where the company has licensed access. What the Times does not move, today, is the consumer answer inside ChatGPT, Perplexity, or Claude.
The bottom line
The Times made a defensible bet. The bet may pay out structurally — through a favorable ruling, mandatory licensing terms, retroactive repricing, and a stronger position for the entire publisher class. It may also leave the Times entering the answer-engine layer as a latecomer to citation patterns that hardened while the lawsuit was pending.
Until then, the Times's archive remains largely outside the answer-engine ecosystem it is challenging in court.
Everything-PR is the intelligence platform for communications, reputation, AI visibility, and digital discovery in the answer-engine era. Publishing since 2009. Covers SEO, GEO, digital PR, and the sources cited by AI answer engines.
