The New York Times, the Daily News, and other publishers allege that OpenAI and Microsoft copied millions of articles to build ChatGPT. In the multidistrict litigation (MDL) in the U.S. District Court for the Southern District of New York, cross-motions for summary judgment and a sanctions motion against OpenAI are pending. The parties filed oppositions on October 23, 2026, and reply briefs are due November 20, 2026.
Latest development Oct 23, 2026
On October 23, 2026, OpenAI, Microsoft, and the news plaintiffs, including The New York Times, filed oppositions to each other’s summary judgment motions in MDL No. 3143. The news plaintiffs argued that OpenAI’s fair use defense wrongly focuses on the technology rather than on substitutive use of their journalism. Reply briefs are due November 20, 2026, under the revised schedule Judge Stein set on September 24, 2026.
Why it matters
The case tests whether copying paywalled journalism to train large language models, and to generate outputs, is fair use. The answer could shape settlements and licensing terms between AI developers and news publishers.
Background
The New York Times Company sued OpenAI and Microsoft Corporation in the U.S. District Court for the Southern District of New York in December 2023. It alleges that millions of Times articles were copied to train the GPT models behind ChatGPT and Copilot. It also alleges that the products reproduce and closely mimic its journalism, including paywalled content.
In March 2025, Judge Sidney H. Stein largely denied the motions to dismiss, preserving the core direct and contributory infringement claims and dismissing the Digital Millennium Copyright Act (DMCA) theories. In April 2025, the Judicial Panel on Multidistrict Litigation centralized the news-publisher and author actions before Judge Stein as In re OpenAI, Inc. Copyright Infringement Litigation (MDL No. 3143). In November 2025, Magistrate Judge Ona T. Wang ordered OpenAI to produce a sample of 20 million de-identified ChatGPT conversation logs. Judge Stein affirmed that order on January 5, 2026, over OpenAI’s user-privacy objections.
On July 9, 2026, the Times, the Daily News, and 15 other media plaintiffs moved for sanctions. They allege that OpenAI falsely told the court it could not search its training corpora and output logs for their works while internally running such searches. They also allege that OpenAI deleted billions of ChatGPT outputs, and they seek adverse findings, exclusion of the log sample, and fees. OpenAI opposed the motion on August 14, 2026. On August 31, 2026, Judge Stein ordered the Times to show cause why its case should not be stayed pending summary judgment in the other active MDL cases.
On September 1, 2026, the Department of Justice (DOJ) filed a statement of interest arguing that training large language models on copyrighted works is generally fair use. On September 4, 2026, OpenAI, Microsoft, and five news organizations cross-moved for summary judgment. The Times argues that the defendants copied its works at scale to build commercial substitutes, while OpenAI argues that existing precedent permits the training. Unredacted portions of the briefs, filed publicly on September 17, 2026, included a 2023 internal Microsoft memo calling AI training on scraped content “an astonishing theft of unprecedented proportions.” Under a revised schedule Judge Stein set on September 24, 2026, the parties filed oppositions on October 23, 2026, and replies are due November 20, 2026. In their oppositions, the news plaintiffs argued that OpenAI’s fair use defense wrongly focuses on the technology rather than on substitutive use of their journalism. Bellwether selection is expected in late 2026.
Timeline 15 events
-
FilingParties file oppositions to summary judgment motions
OpenAI, Microsoft and the news plaintiffs filed oppositions to each other’s summary judgment motions. The news plaintiffs argued OpenAI’s fair use defense wrongly focuses on the technology rather than on substitutive use of their journalism.
OpenAI, Microsoft's summary judgment arguments fail, US news outlets sayThird Circuit ROSS ruling peppers news outlets' opposition to ChatGPT fair use defense
-
FilingUSA Today Co. newspapers sue OpenAI in related New York action
USA Today Co. and about 18 affiliated newspapers alleged that OpenAI used their articles without authorization to train GPT models and sought more than $250M. The suit is related to, but separate from, the Times case and MDL No. 3143.
USA Today Adds To Pile Of Newspaper Suits Against OpenAI - Law360 UKUSA Today network, including Courier Journal, sue OpenAIUSA Today Co. sues OpenAI for more than $250M over AI training dataUSA Today sues OpenAI in yet another tag-along copyright suit
-
OrderJudge Stein resets summary judgment briefing schedule
Judge Stein issued a revised schedule for the pending cross-motions for summary judgment, setting oppositions for October 23, amicus deadlines for October 30, and replies for November 20, 2026.
OpenAI Copyright Infringement - Center for AI and Digital PolicyOpenAI, Microsoft's summary judgment arguments fail, US news outlets say
-
FilingUnsealed summary judgment briefs reveal internal Microsoft memo on scraping
In a 2023 memo, a Microsoft executive called AI training on scraped content “an astonishing theft of unprecedented proportions.” The filings also showed more than 91,692 copies of Times, Daily News, and Center for Investigative Reporting works in OpenAI’s training datasets.
Microsoft exec called AI scraping 'the largest theft of labor in human history,' new unredacted filings revealMicrosoft exec called AI scraping 'the largest theft of labor in human history,' new unredacted filings revealOpenAI and Microsoft Internal Documents Describe News as an Unpriced Input — and Litigation Is Filling the Gap
-
FilingParties file cross-motions for summary judgment
OpenAI, Microsoft, and five news organizations each moved before Judge Stein. The Times argued that the defendants copied its works at scale to build commercial substitutes, and OpenAI argued that existing copyright precedent permits the training.
OpenAI and Microsoft ask judge to end 10.8 million-article copyright caseHistoric NYT v. OpenAI copyright battle heats upJudge Stein issues stipulated sealing order as briefing for summary judgment commences in OpenAI MDL suits
-
FilingDOJ files statement of interest backing OpenAI’s fair use defense
The Department of Justice’s 20-page brief argued that training large language models on copyrighted works is generally fair use. It argued that restricting AI training would harm U.S. scientific progress, economic competitiveness, and national security.
DOJ Sides with OpenAI, Warns Obstacles to AI Development Threaten National SecurityDOJ urges judge to rule for OpenAI, Microsoft in N.Y. Times lawsuit
-
OrderJudge Stein orders Times to show cause on staying its case
The order asked why the Times case should not be stayed pending summary judgment in other MDL cases. The Times had until September 11, and the defendants until September 18, to respond.
-
FilingOpenAI opposes news plaintiffs’ sanctions motion
OpenAI urged the court to reject sanctions sought by the New York Times and other news organizations, arguing it did not violate discovery obligations regarding the 20-million-chat log sample.
-
OtherOpenAI publicly denies sanctions allegations
OpenAI spokesperson Drew Pusateri denied the allegations in response to the publishers’ sanctions motion. He said the Times was making “blatantly false allegations” as its case weakened and that OpenAI would continue defending user privacy and fair use principles.
-
FilingNews plaintiffs move for sanctions against OpenAI
Seventeen media plaintiffs alleged that OpenAI concealed its ability to search training data and output logs and deleted billions of ChatGPT outputs. They sought adverse findings and fees.
-
OrderJudge Stein affirms the 20-million-log order
The court overruled OpenAI’s user-privacy objections and held that de-identification and the protective order were sufficient safeguards.
-
OrderMagistrate Judge Wang orders production of 20 million ChatGPT logs
The court rejected OpenAI’s proposal to run targeted keyword searches and produce only conversations implicating the plaintiffs’ works.
-
OrderJPML centralizes the litigation as MDL No. 3143
The panel consolidated 12 news-publisher and author actions against OpenAI and Microsoft before Judge Stein in the Southern District of New York. More actions followed.
-
OrderJudge Stein largely denies motions to dismiss
Core direct and contributory copyright claims survived, and the Digital Millennium Copyright Act (DMCA) claims were dismissed. The full opinion issued on April 4, 2025.
-
FilingNew York Times sues OpenAI and Microsoft
The Times alleged that millions of its articles were used without authorization to train GPT models and that ChatGPT reproduces its journalism.
Experimental AI forecast
How it is madeHow will Judge Stein resolve the fair use summary judgment motions in the news-publisher cases in MDL No. 3143 by December 31, 2027?
- Split ruling or fair use left for trial35%
- Still pending on December 31, 202730%
- OpenAI and Microsoft win on fair use13%
- Times settles with OpenAI and Microsoft13%
- Court rejects OpenAI’s fair use defense9%
- Resolves by
- Dec 31, 2027in 14 months
- Record confidence
- Medium
A split ruling, or a denial that leaves fair use for trial, is the most likely result, at 35% by December 31, 2027. Fair use is fact-bound, and the record mixes training claims with output and substitution claims that courts tend to treat differently.
How it resolves. Resolves on the first public order by Judge Stein deciding the fair use issue on the pending cross-motions for summary judgment in the news-publisher cases in MDL No. 3143. “OpenAI and Microsoft win” means the court holds that fair use bars the news plaintiffs’ infringement claims in full. “Court rejects fair use” means the court grants the news plaintiffs summary judgment that fair use does not apply to the challenged training and output uses. “Split ruling” means any other ruling on the motions, including partial grants or denial of both sides’ motions so fair use goes to trial. “Times settles” means the Times publicly announces a settlement with OpenAI, or with both defendants, before any such ruling. “Still pending” means none of these has occurred by December 31, 2027, including because the Times case is stayed.
Reasoning
Starting point. Federal district court summary judgment rulings on fair use in generative AI training copyright cases: Of the few U.S. rulings so far (Thomson Reuters v. Ross, Bartz v. Anthropic, Kadrey v. Meta), roughly two of three were split or narrow rather than clean wins for either side.
Comparable AI training cases have mostly produced divided or narrow fair use rulings rather than clean wins, so the base rate favors a split outcome. Here the Times presses both training and output claims, and evidence of verbatim copies and a pending sanctions motion make a full defense win harder. The DOJ statement of interest and the training-focused precedent support OpenAI on the training claims, which keeps a defense win plausible at 13%. A full plaintiff win is the least likely, since courts rarely reject fair use as a matter of law at this stage. Briefing closes November 20, 2026, but cross-motions in a large MDL, plus the sanctions motion, can take many months to decide. The August 31, 2026, show-cause order on staying the Times case adds real risk that its motion waits behind the other cases, which keeps “still pending” at 30%. Settlement stays at 13% because publisher licensing deals are common, but the Times has litigated for 34.2 months and seeks precedent. Nothing in the record since the October 10, 2026, forecast moved any outcome by five points or more. The oppositions were filed on schedule, which slightly lowers the delay risk.
Fair use is fact-bound and often split training versus outputs · Possible stay of the Times case pending other MDL rulings · Pending sanctions motion over alleged concealment and deleted outputs · DOJ statement of interest supporting fair use for training · Industry licensing deals create settlement pressure
Generated automatically by Claude Opus 5.5 on Nov 1, 2026 from this case’s record on this site, and published without editorial review. The probabilities are the model’s judgment, not measured frequencies, and they are scored publicly as cases resolve. Not legal or investment advice. How the forecasts are made
Disclosure: Anthropic, which makes the model that wrote this forecast, competes with or has commercial ties to a party to this case.