NYT v. Microsoft, OpenAI et al.

NYT v. Microsoft, OpenAI et al. is a copyright suit filed by The New York Times Company against Microsoft and OpenAI in the U.S. District Court for the Southern District of New York, the first major US news-publisher copyright case against a frontier AI developer. Core copyright claims survived a motion to dismiss in April 2025, and the case is consolidated within a multidistrict litigation (MDL) of news-organization plaintiffs. Summary-judgment briefing closed in April 2026, and in July 2026 the publisher plaintiffs moved to sanction OpenAI over alleged withholding and destruction of evidence.

Background

The New York Times Company alleges that millions of Times articles were ingested for training, including via overweighted Common Crawl segments. Exhibit J to the complaint sets out roughly 100 examples of GPT-4 and Bing Chat producing near-verbatim Times text, including the 2019 Pulitzer-winning NYC taxi-lending investigation, the "iEconomy" Apple series, paywalled recent articles, and Wirecutter recommendations. The complaint also alleges hallucinated attributions — fabricated "quotes" and Wirecutter picks falsely ascribed to the Times — which ground the trademark-dilution claim.

The case is the first major US news-publisher copyright suit against a frontier AI lab, and it has framed the doctrinal approach for subsequent publisher suits, including those brought by Alden, the Daily News, the Intercept, and the Center for Investigative Reporting. The survival of core claims at the motion-to-dismiss stage indicated that a theory combining "memorization" with a "substitutive market effect" is a viable basis for AI copyright liability, working against a defense resting purely on transformative use. The case serves as the anchor case in AI Copyright Litigation — Analysis.

Claims

The complaint pleads six causes of action:

1. Direct copyright infringement (17 U.S.C. § 501) — against Microsoft and OpenAI.
2. Vicarious copyright infringement — against Microsoft.
3. Contributory copyright infringement — against Microsoft and OpenAI.
4. DMCA § 1202 — removal of copyright management information (CMI).
5. Common-law unfair competition by misappropriation (hot-news doctrine).
6. Trademark dilution (15 U.S.C. § 1125(c)) — based on hallucinated outputs falsely attributed to the Times.

Relief sought

The Times seeks statutory damages of up to $150,000 per willful infringement, alleged to total "billions" in the aggregate; injunctive relief including destruction of infringing datasets and models; and attorneys' fees.

Procedural history

In April 2025, the core copyright claims survived OpenAI's and Microsoft's motion to dismiss. Some secondary claims were trimmed, while the primary infringement and DMCA counts proceeded.

On January 5, 2026, District Judge Sidney Stein affirmed Magistrate Judge Ona Wang's order compelling OpenAI to produce 20 million anonymized ChatGPT logs, about 0.5% of preserved logs, across the consolidated MDL, over OpenAI's privacy and reconsideration objections. The court held that logs across a broad range of queries are discoverable because they bear on OpenAI's fair-use defense — specifically, whether ChatGPT outputs compete with or substitute for copyrighted works. The order covers the news-organization plaintiffs within the MDL, including NYT, the Chicago Tribune, and others (Source: news.bloomberglaw.com; natlawreview.com).

The discovery order was the first instance of a court compelling large-scale production of user-generated AI outputs, as opposed to training data alone, in an AI copyright case. The 20-million-log dataset is described as the empirical foundation for the "memorization + substitutive market effect" theory: plaintiffs must show that ChatGPT outputs, in the wild, actually substitute for copyrighted works, and the breadth of the production order means OpenAI cannot limit discovery to logs already known to contain plaintiffs' material. This shifts the regurgitation question from the anecdotal basis of Exhibit J toward a statistical one.

Discovery continued through spring 2026. In March 2026, the Times won a bid for a diary kept by OpenAI's president in connection with the consolidated copyright suits, and OpenAI was separately ordered to produce testimony from the Musk v. Altman litigation (Source: news.bloomberglaw.com). Expert reports were due November 14, 2025, and summary-judgment briefing closed on April 2, 2026 under the scheduling order; on April 8, 2026, Magistrate Judge Wang granted the news-organization and consolidated author plaintiffs additional deposition time after finding OpenAI's Rule 30(b)(6) corporate designee unprepared on noticed topics, ordering OpenAI to produce a properly prepared witness (Source: ailawsuittracker.com).

Sanctions motion (July 2026)

On July 9, 2026, a group of publisher plaintiffs led by the Times and the New York Daily News asked the court to sanction OpenAI, alleging it withheld and destroyed key evidence in the copyright dispute (Source: reuters.com). The publishers allege OpenAI concealed internal tools and datasets that could identify copyrighted journalism in ChatGPT outputs (Source: techcrunch.com). Daily News attorney Steven Lieberman said the motion "asks the court to punish OpenAI for hiding and destroying evidence" (Source: reuters.com). Per Ars Technica's account of the filing, the plaintiffs allege OpenAI concealed for two years its ability to search ChatGPT logs — including undisclosed samples of 10 million and 78 million logs — applied 19 billion AI-generated redactions to the sample it did produce, and deleted logs despite a preservation order; OpenAI called the allegations "blatantly false" (Source: arstechnica.com). The motion's resolution was pending as of July 10, 2026.

Statement of interest of the United States (September 2026)

On September 1, 2026 the Department of Justice filed a statement of interest under 28 U.S.C. § 517 in the consolidated MDL, docketed at 1:25-cv-03483-SHS-OTW as Document 316 and marked as relating to all matters (Statement of Interest of the United States — In re OpenAI, Inc. Copyright Infringement Litigation (September 1, 2026)). It is signed by Associate Attorney General Stanley E. Woodward, Jr., Assistant Attorney General Brett Shumate, and Senior Counsel Michael Weisbuch, and states that its arguments apply to "all parties in this litigation and the related cases, including book authors and publishers."

The United States argues that copying works to train a large language model is fair use. Its organizing principle is the use-by-use requirement of Andy Warhol Foundation v. Goldsmith: training and output are distinct uses requiring separate analysis, so that "whatever legal questions certain output uses might raise, that should not bear on the transformative nature (or any other aspect) of the training use." On the first factor it calls training "extraordinarily transformative," borrowing "spectacularly so" from *Bartz v. Anthropic*. On the fourth it argues that training "does not reveal anything to the public at all," so cannot produce the substitutive competition the factor requires, and that outputs which copy no protected elements cannot cause cognizable harm merely by occupying the same genre.

The filing devotes a section to criticizing the market-dilution reasoning in *Kadrey v. Meta Platforms* as "[c]ontrary dicta" that "improperly collapsed LLM training and LLM outputs into a single continuous use." A footnote states that the Register of Copyrights, "who is currently challenging her removal," endorsed a similar theory in the Copyright Office's May 2025 Part 3 report, that her understanding "does not warrant deference" under Loper Bright, and that her reasoning was "threadbare."

Two further arguments sit outside the fair-use factors. The United States asserts that an adverse ruling "would hamper competition in the market for LLMs, because only the largest technology companies might have the capital necessary to pay licensing fees," and that such fees "function primarily as large subsidies for old mainstream media companies" — while a footnote takes no position on whether a licensing regime would be feasible, and notes that publishers have entered licensing agreements regardless of the fair-use question. And it grounds the government's interest in Executive Order 14179, the June 2, 2026 executive order on advanced AI innovation and security, and the National Policy Framework legislative recommendations of March 2026.

A statement of interest is advisory. The United States is not a party, and the filing states that it does not contend any conduct at issue was authorized by or for the benefit of the government.

The filing surprised the U.S. Patent and Trademark Office and the Copyright Office, according to sources cited by Axios on September 19, 2026, which reported that the government's position had not been coordinated among the bodies that work closely on the issue and that, unlike many statements of interest, no career antitrust attorneys signed it alongside the senior Justice Department officials. The same account noted that President Trump had removed the Copyright Office's director, Shira Perlmutter, in 2025 and that her challenge to the removal was ongoing. Ziff Davis chief executive Vivek Shah had argued in an op-ed that the government's reasoning was muddled, and Axios observed that the brief calls licensing enforcement cumbersome without taking a position on whether a licensing regime is feasible. The USPTO and the Copyright Office declined to comment and the Justice Department did not respond (Source: axios.com).

On September 4, 2026 the publishers in the consolidated case before Judge Stein moved for summary judgment, as did OpenAI and Microsoft (Source: insideaipolicy.com; Source: geekwire.com).

Microsoft filed separate briefs the same day in the News Cases and in the Books Plaintiffs' cases. Reporting on the News Cases brief states that of 8.2 million Copilot chat logs Microsoft produced in discovery — logs Microsoft says were selected "because they hit on keywords implicating use of News Plaintiffs' websites, and therefore the most likely to contain News Plaintiffs' works" — 59,545 contained at least 16 words in common with news content used to ground the model, and that an expert for the Center for Investigative Reporting found 51 instances of substantial overlap in that dataset. Ian Crosby, lead counsel for The New York Times, responded that discovery leads "to only one conclusion: Microsoft and OpenAI stole from The New York Times to make commercial products that substitute for its journalism" (Source: theverge.com). The parallel brief in the Books Plaintiffs' cases, which is on the public docket, gives a different set of figures over the same 8.2 million conversations — 24 responses containing 30 matching words, and no regurgitation found for 202 of 212 asserted works — attributing the review to the Books Plaintiffs' expert Dr. Shawn Shan (Microsoft's Memorandum in Support of Summary Judgment, Books Plaintiffs' Consolidated Cases (S.D.N.Y., September 2026)); the two briefs use different match thresholds and different expert reports, and the figures are not interchangeable. The books brief expressly reserves the news question, stating that "every LLM-based tool must be considered in light of its own design and function" and pointing to the separate News Cases brief.

Coverage of the completed briefing on September 8, 2026 described the parties as taking sharply different positions on how far model outputs resembled the training material, a question bearing on the fourth fair-use factor. The Times argued that the companies copied its works at scale to build commercial substitutes and that the case should be measured against Supreme Court and Second Circuit precedent rather than the recent California district decisions in *Kadrey v. Meta* and *Bartz v. Anthropic* that OpenAI cited. Judge Stein is expected to rule in the coming months on whether the case, or parts of it, proceeds to trial in 2027 (Source: axios.com). Trade coverage the following day framed the dispute as turning on the "amount and substantiality" factor; only the lede was retrievable, so its reasoning is not established here (Source: insideaipolicy.com). The Axios account dates the Justice Department's supporting brief to September 3, 2026; the docketed statement of interest described above is dated September 1, and the difference is not resolved here.

Unredacted summary-judgment material (September 2026)

Unredacted portions of the publishers' summary-judgment filings became public on September 17, 2026. Much of the material comes from the Times's own brief rather than from the underlying exhibits, which remain sealed, and the quotations reached the record without their original context. According to the filing:

• In a January 2023 internal memo, Brent Hecht, Microsoft's director of applied science, called AI scraping "an astonishing theft of unprecedented proportions" and "the largest theft of labor in human history." A January 2024 Microsoft presentation he wrote described a decline of up to 93% in click-through rates to the Times's domain from Copilot's answer engine, compared with traditional Bing search, as a "doom loop" that would "hurt the performance of our models and the entire web at the same time."
• Nick Turley, OpenAI's head of ChatGPT, wrote internally that publishers face an "existential threat" from products like ChatGPT, which are "largely substitutive" and "will get more and more substitutive as they get better"; OpenAI president Greg Brockman described the models as "excellent at news," and replied "ah nice" when researcher Nick Ryder told him of a "hack to get around nytimes paywall."
• Satya Nadella testified in a deposition earlier in 2026 that "anything that is paywalled should be licensed by anyone who wants to use it…for grounding or training," and that had he known OpenAI had scraped and trained on paywalled information he would have invoked Microsoft's right to require OpenAI to retrain its models; he also agreed that conversing with chatbots "has substituted" for going to the underlying source.
• OpenAI's mid-training datasets contain more than 91,692 copies of works published by the Times, the Daily News and the Center for Investigative Reporting, and a Common Crawl-derived dataset included more than 2 million documents from nytimes.com. "OpenAI delivered the entire GPT-3 training dataset to Microsoft," and Microsoft "provided training data to OpenAI through initiatives called Project Taxi and Project Mango"; the Project Mango data was assembled into a dataset containing copies of at least 160,903 unique works from the news publishers.
• OpenAI researchers stripped copyright notices from training data because they "wouldn't want model outputting" "copyright notices" to users.

OpenAI and Microsoft did not respond to requests for comment (Source: techcrunch.com). The substitution statements bear on the fourth fair-use factor at issue in the cross-motions for summary judgment described above.

Source summary

A full case summary with claims and exhibits is in NYT v. OpenAI and Microsoft — Complaint (Dec 2023) (sources/).

Related filings outside the consolidated proceeding

A further publisher suit was filed in the same district on September 4, 2026 by The Seattle Times Company and Newsday LLC, No. 1:26-cv-07644, against nine OpenAI entities and Microsoft. Its complaint pleads that the defendants consented to jurisdiction and venue in this district in NYT v. Microsoft, Raw Story Media v. OpenAI, The Intercept Media v. OpenAI and Daily News v. Microsoft, all consolidated here as No. 25-md-03143. Whether the new case is transferred into the multidistrict proceeding is not resolved on the docket as of September 5, 2026 (The Seattle Times Co. and Newsday LLC v. OpenAI and Microsoft).

Relationships

• supports: AI Copyright, AI Copyright Litigation — Analysis, Training Data Walls, AB 2013 — Training Data Documentation (California).
• depends-on: AI Copyright.
• related: OpenAI, Microsoft, AI in Journalism and Media, United States Copyright Office.
• contradicts: Statement of Interest of the United States — In re OpenAI, Inc. Copyright Infringement Litigation (September 1, 2026) — the United States argues for the fair-use position the Times opposes.
• related: Microsoft's Memorandum in Support of Summary Judgment, Books Plaintiffs' Consolidated Cases (S.D.N.Y., September 2026) — Microsoft's parallel summary-judgment brief in the Books Plaintiffs' cases.
• related: Authors Guild v. OpenAI