Thirty-five U.S. local and regional newspaper publishers filed suit in federal court on June 24, 2026, accusing Microsoft and multiple OpenAI entities of systematically misappropriating copyrighted journalism to train and operate ChatGPT and Microsoft Copilot. The complaint, lodged in the Southern District of New York, marks one of the most coordinated legal actions to date from the newspaper industry against the generative AI giants and expands a legal front that already includes The New York Times and other national outlets.

The lawsuit argues that Microsoft and OpenAI copied vast amounts of publisher content without permission, stripped out copyright management information in violation of the Digital Millennium Copyright Act, and now deploy that material inside commercial products that directly compete with the publishers’ own websites and subscription services. Because Microsoft Copilot is deeply embedded across Windows 11, Microsoft Edge, and the company’s cloud productivity stack, the case rapidly becomes a Windows story—potentially affecting millions of users who encounter AI-summarized news every day.

What the Publishers Are Claiming

According to the publishers’ legal filing, the core allegation mirrors the narrative that has played out in similar lawsuits since late 2023: that OpenAI and Microsoft used “shadow library” datasets containing copyrighted news articles to train large language models, including the GPT family that underpins both ChatGPT and Copilot. The complaint asserts that the companies never sought licensing agreements or provided compensation, effectively freeloading on the substantial investments local newspapers make in original reporting.

The DMCA claims add a specific technical dimension. The publishers argue that when the AI models reproduce article excerpts or close paraphrases, they routinely omit bylines, publication dates, hyperlinks, and other copyright management information that federal law requires be preserved. By doing so, the output creates the false impression that the content is AI-generated or uncopyrighted, destroying critical attribution and traffic that local newsrooms depend on. The suit demands statutory damages under the DMCA, which can reach up to $25,000 per violation, as well as an injunction to stop further use of the publishers’ content.

Other counts in the complaint include direct and contributory copyright infringement, unfair competition, and unjust enrichment. The publishers seek not only monetary relief but an order requiring Microsoft and OpenAI to delete all training data that was scraped from their publications and to implement technological safeguards preventing future use.

Why Local Newspapers Are Joining the Fight

The group of thirty-five plaintiffs represents a cross-section of small and mid-sized American news organizations—community dailies, alt-weeklies, and regional chains that have already been battered by two decades of digital disruption. Unlike The New York Times or The Wall Street Journal, these outlets rarely have the budget to mount armies of lawyers. The coordinated filing suggests they are pooling resources, possibly through a newspaper trade association or a litigation funder.

For these publishers, the stakes are existential. If ChatGPT and Copilot can reliably summarize a city council meeting or a high school football game—information originally gathered by a local reporter—the incentive for readers to visit the newspaper’s site evaporates. Subscription revenue, already fragile, could crater. Advertising, tied to page views, follows. The publishers describe a future where AI becomes a “middleman” that replaces direct reader relationships, extracting value while starving the very journalism that feeds it.

In a statement accompanying the filing, the plaintiffs highlighted that Microsoft Copilot is now the default AI assistant on more than 400 million Windows 11 devices. “By design, Copilot encourages users to bypass publishers entirely,” the statement reads. “Every query answered with scraped news content is a billable interaction for Microsoft and a lost opportunity for the newspaper that actually reported the story.”

Microsoft Copilot’s Integration Deepens the Threat

Windows 11 version 24H2 shipped with Copilot pinned to the taskbar by default, and recent updates have woven the assistant into File Explorer, Edge, and the system-wide search interface. Copilot can generate summaries, answer questions, and even draft content in real time—often pulling information from live web results that include newspaper articles. Users can ask Copilot “what happened at last night’s school board meeting,” and the assistant may synthesize a response that closely tracks the publisher’s reporting without ever directing the user to the original source.

Microsoft has promoted Copilot as a productivity tool, but the publishers’ lawsuit paints it as a content parasite. The complaint cites specific examples where Copilot reproduced near-verbatim passages from subscriber-only stories after a user asked a tailored question, effectively unlocking paywalled material. The publishers argue this undermines their business model and violates the terms of their websites, which explicitly restrict scraping and use of content for AI training.

The lawsuit also names Microsoft’s enterprise AI offerings, including Copilot for Microsoft 365 and Azure OpenAI Service, arguing that Microsoft profits when business customers use these tools to generate reports, emails, or market analyses that incorporate copyrighted news content. The publishers claim Microsoft is aware of the infringement because its own documentation warns enterprise customers to check for copyrighted output, yet the company has not implemented filtering to prevent it.

This case is the latest in a cascade of lawsuits that began in earnest when The New York Times sued OpenAI and Microsoft in December 2023. Since then, groups of authors, visual artists, music publishers, and coders have all filed class-action complaints challenging the legality of AI training on copyrighted works. The cases raise fundamental questions: Does ingesting copyrighted content for training constitute fair use? When an AI model generates output that resembles the input, who is liable? And what does the DMCA require of AI companies that remove metadata from training data?

Courts have been slow to issue definitive rulings. In mid-2025, a federal judge in the Southern District of New York allowed the core copyright claims in the Authors Guild suit against OpenAI to proceed, rejecting a motion to dismiss on fair use grounds. That ruling signaled that plaintiffs have a viable path, but no jury has yet decided a full trial. The newspaper cases add urgency because journalism is often cited as a public good, and courts may be more sympathetic to the argument that AI threatens the free press.

Microsoft and OpenAI have consistently argued that training on publicly available data is protected by fair use, comparing it to a human reading a newspaper and learning facts. In a February 2025 blog post, OpenAI stated that “training AI models using publicly available internet materials is a transformative use that does not compete with the original works.” Microsoft, in its 2025 annual report, described copyright lawsuits as “without merit” but acknowledged they could result in “significant costs” if plaintiffs prevail.

The DMCA claims add a novel twist. If the plaintiff’s argument that AI training systematically strips copyright management information holds, the statutory damages could run into staggering sums. Even a sliver of the alleged violations could total hundreds of millions of dollars. That threat alone may pressure companies to settle or to introduce more robust CMI preservation in training pipelines.

Local Newsrooms at an Inflection Point

The lawsuit arrives as local journalism continues to shrink, with an average of two newspapers closing per week in the United States during 2025, according to the Medill Local News Initiative. The publishers note that while national outlets have been able to strike licensing deals with AI companies—The Associated Press, Axel Springer, and others have signed multi-year agreements—local publishers have been largely ignored. The complaint asserts that Microsoft and OpenAI have deliberately focused on scraping local content because it is less likely to be locked behind sophisticated paywalls and more vulnerable to automated collection.

Some publishers have experimented with AI themselves, using tools to automate sports box scores or weather reports, but those efforts typically rely on original data collection. The lawsuit stresses that these publishers are not anti-technology; they are anti-theft. Several plaintiffs say they would welcome a licensing framework that pays newspapers a fair market rate for content used in AI training and output, but no such offer has materialized.

The economic math is stark. A single investigative story can cost a local paper $5,000 or more in reporter time, legal review, and editing. If Copilot can instantly distill that story into a free answer, the newspaper’s return on investment evaporates. Multiply that by hundreds of stories across thirty-five publishers, and the total alleged damages quickly climb into nine figures.

What Microsoft and OpenAI Might Argue

In previous cases, the tech companies have leaned on several defenses. They argue that copyright protects expression, not facts, and that AI models learn the factual information from articles rather than reproducing the expression. They also invoke the intermediary safe harbors of Section 230 of the Communications Decency Act, claiming that when Copilot surfaces web content, it is acting as a neutral platform. However, recent court decisions have narrowed Section 230’s scope for algorithmic recommendations, and the direct integration of Copilot into Windows may weaken that argument.

On the DMCA front, Microsoft and OpenAI may claim that the removal of CMI is not intentional and occurs as a byproduct of the training process, not as a deliberate scheme. They could also assert that any CMI removal falls under an exemption for research or that the publishers’ metadata was insufficiently robust to trigger DMCA protection.

A more pragmatic defense is that licensing deals are already happening. Microsoft announced in 2025 that it had inked agreements with several hundred news organizations globally. But that program, called NewsGuard AI, covers only a fraction of U.S. local newspapers, and none of the current plaintiffs are participants. The lawsuit may put pressure on Microsoft to expand the program rapidly or face broader liability.

Implications for Windows Users

For everyday Windows users, the immediate effect is unlikely to change their daily interaction with Copilot—yet. But if the publishers win an injunction, Microsoft could be forced to modify how Copilot retrieves and presents news. That might mean stricter limits on the verbatim reproduction of article text, more prominent source attribution, or a “licensing wall” that blocks certain queries unless the publisher has been compensated. For users, the experience could become less seamless or more paywalled.

The case could also accelerate Microsoft’s move toward a premium Copilot tier that shares revenue with content creators. Microsoft has tested a Copilot Pro subscription that includes citation features and deeper integration with its Bing search index; a settlement might require a slice of that revenue to flow to publishers. Alternatively, if Microsoft loses at trial, it could decide to pull Copilot entirely from certain markets or scale back its news summarization features.

Enterprise customers should watch closely too. A decision that AI training on copyrighted content requires a license would reshape the enterprise AI market, potentially raising costs and slowing deployment. Microsoft’s Copilot for Microsoft 365 depends on the same underlying models; a court-imposed licensing regime could ripple into every corner of the Office ecosystem.

The Southern District of New York has become ground zero for AI copyright litigation, with several cases now progressing through discovery. This local newspaper suit, while not the first, may be among the most consequential because it represents a class of plaintiffs that judges and juries might instinctively root for—community institutions that have been hollowed out by the same platform economy that enriched big tech.

No trial date has been set, and both Microsoft and OpenAI are expected to file motions to dismiss by late summer 2026. A ruling on those motions could come by early 2027. In the meantime, the case will generate thousands of pages of discovery, including internal emails about how OpenAI assembled its training datasets and what Microsoft engineers knew about the data’s provenance. That discovery alone could prove explosive, potentially revealing practices that go beyond what is currently alleged.

For now, the 35 publishers have fired a shot that resonates far beyond the newsrooms of their small towns. The lawsuit asks a fundamental question that Congress, the courts, and the public have yet to answer: can the foundational technology of the next computing era be built on the uncompensated labor of the last century’s information producers?