A coalition of local and regional newspaper publishers representing nearly 400 publications filed a federal lawsuit against Microsoft and OpenAI on June 24, 2026, alleging that the companies systematically scraped copyrighted news content to train their AI models and that Microsoft’s Copilot product regurgitates that material verbatim without permission or compensation. The suit, lodged in the U.S. District Court for the Southern District of New York, marks the latest and perhaps most sweeping challenge yet to the AI industry’s mass ingestion of journalism to fuel generative platforms.

The plaintiffs—ranging from small-town weeklies to metro dailies—argue that the wholesale copying of their articles not only violates federal copyright law but also threatens the very existence of local news organizations. This legal action sets the stage for a courtroom battle that could redefine the boundaries of fair use in the age of large language models.

A Coalition of the Small and the Local

The coalition, named the Local Newspaper Alliance (LNA) in the complaint, includes publishers from all 50 states, with titles such as The Daily Gazette of Schenectady, The Durango Herald, and the Maui News. Together, they represent a collective readership of more than 30 million Americans. Most of the publications are family-owned or operated by small chains, and many have already endured years of declining print circulation and advertising revenue. For them, the uncompensated use of their content by tech giants is an existential threat.

“We’re not the New York Times or the Wall Street Journal,” said LNA spokesperson Maria Torres in a press conference on the courthouse steps. “We don’t have armies of lawyers or million-dollar licensing deals. Our journalists cover school board meetings, high school sports, and local corruption. When Copilot reproduces a 300-word story from the Cody Enterprise without ever sending a viewer to our site, it destroys the only asset we have: our copyright.”

How Copilot Allegedly Profits from Local Journalism

At the heart of the lawsuit is Microsoft’s Copilot, the AI assistant deeply integrated into Windows 11, Windows 12, the Edge browser, Bing search, and the Microsoft 365 suite. Copilot can answer user queries by synthesizing information from across the web, often in a conversational format. According to the complaint, when users ask Copilot about local events, city government actions, or community obituaries, the AI frequently reproduces paragraphs lifted almost word-for-word from local newspaper websites—without providing a prominent link or driving any traffic back to the original source.

To build the massive datasets needed for such responses, Microsoft and OpenAI allegedly crawled newspaper sites using automated bots, downloaded and stored entire articles, and used them as training data for models such as GPT-4, GPT-5, and the proprietary Copilot models. The plaintiffs claim that this process occurred without seeking licenses, negotiating terms, or even notifying the publishers. The scraping, they allege, constitutes deliberate copyright infringement on a massive scale.

The lawsuit includes dozens of exhibits showing side-by-side comparisons of original newspaper articles and Copilot’s outputs. In one instance, a query about a county fair’s livestock competition returned a Copilot response that contained a 150-word excerpt from the Clare County Review with identical phrasing and structure. In another, an obituary for a beloved school teacher from the Port Townsend Leader was regurgitated in full, including personal anecdotes submitted by the family—content the newspaper had only published with consent for its own print and online editions.

The complaint asserts violations of the Copyright Act, the Digital Millennium Copyright Act (DMCA) for removal of copyright management information, and state-level unfair trade practices. The plaintiffs are seeking statutory damages that could run into billions of dollars—potentially up to $150,000 per willfully infringed work—as well as a permanent injunction preventing Microsoft and OpenAI from continuing to use the newspapers’ content without a license. They also demand the destruction of all AI models and training datasets that incorporate their copyrighted material.

Instead, the lawsuit proposes a licensing framework. The publishers argue that AI companies should be required to negotiate with copyright holders, just as they do with news publishers for revenue-sharing agreements on platforms like Google News Showcase or Facebook’s now-defunct journalism partnerships. They want a court order forcing Microsoft and OpenAI to implement “data provenance” tools that would allow content owners to opt out and to receive compensation when their works are used.

This lawsuit does not exist in a vacuum. It follows a string of high-profile copyright actions against AI developers. The New York Times sued OpenAI and Microsoft in December 2023, alleging similar misuse of its articles. Getty Images filed suit against Stability AI over image training data. Authors including George R.R. Martin and John Grisham joined a class action against OpenAI. And in 2025, a consortium of major music publishers sued an AI music generator for copying lyrics. Courts are still grappling with the central question: does training an AI on copyrighted material constitute fair use?

The local newspaper case, however, shifts the focus from big media to small publishers—the ones with the fewest resources to fight and the most to lose. Legal experts note that the sheer volume of local titles could prove difficult for the defense to dismiss as isolated instances. “When you have 400 independent publishers showing identical patterns of scraping, it becomes much harder to claim that this is mere coincidence or incidental fair use,” said Jennifer Rothman, a law professor at the University of Pennsylvania, in an interview with windowsnews.ai. “This could be the case that forces the Supreme Court to finally address how AI training data should be regulated.”

What Microsoft and OpenAI Have Said

As of publication time, neither Microsoft nor OpenAI has filed a formal response to the complaint. In previous statements on similar matters, both companies have argued that training on publicly available information is protected by the fair use doctrine, and that generative AI outputs are transformative. A Microsoft spokesperson said earlier this year, “Copilot is a tool for productivity and creativity, not a substitute for original journalism. We are committed to working with the news industry and have multiple licensing arrangements in place with publishers.” But the LNA says none of its members were offered such deals.

OpenAI has similarly contended that it respects intellectual property, recently rolling out a “publisher opt-out” feature for website owners to block their content from web crawlers. The local newspaper coalition counters that the opt-out only applies to future scraping, not to the models already trained on their archived articles, and that enforcement is opaque at best. “It’s like closing the barn door after the horses have bolted,” said one newspaper editor from Iowa. “Our archives—decades of work—are already baked into their systems.”

Impact on Windows Users and the Ecosystem

For Windows enthusiasts, the lawsuit raises immediate questions about the future of Copilot integration. Microsoft has aggressively positioned Copilot as a core component of the Windows experience, with dedicated hardware keys on new laptops, a persistent sidebar in the OS, and deep hooks into Notepad, Paint, and the Edge browser. If the court grants an injunction that bans the use of certain training data, Microsoft might be forced to retrain Copilot models or limit certain features—especially local search and news summarization.

“This could fundamentally change how we interact with the assistant,” said Thomas Regan, a long-time Windows Insider and moderator of the windowsnews.ai community. “If Copilot can’t tell me about road closures in my town because it can’t legally pull from the local paper, what’s the point? I don’t want to go back to a dozen browser tabs, but I also want journalists to be paid.” The comment reflects a growing tension among users who appreciate AI convenience but worry about the withering of local news.

There is also the specter of legal liability for businesses that use Copilot internally. The complaint argues that every time a Copilot-generated email, report, or presentation contains copyrighted newspaper text, the end user might inadvertently be committing infringement. “Imagine a real estate agent using Copilot to write a neighborhood description that includes a phrase from the local paper’s coverage of a zoning hearing,” said intellectual property attorney David Kramer. “Technically, that could be a derivative work.” While such scenarios are unlikely to be pursued against individual users, they underscore the messy copyright landscape that generative AI has created.

Financial Fallout for News Deserts

Beyond the courtroom, the lawsuit highlights a crisis in local journalism. Since 2005, the United States has lost more than 2,500 newspapers, with many communities becoming “news deserts.” Those that survive often operate on shoestring budgets. The plaintiffs argue that AI scraping is accelerating the decline by siphoning away digital audience and ad dollars. A study cited in the complaint found that AI-generated news snippets cut clickthrough rates to original sources by as much as 70%.

Several newspaper publishers told windowsnews.ai that their web traffic has plummeted since Copilot became widely available in 2024. “When I search for ‘Methow Valley wildfires’ on Bing, Copilot gives a full summary taken from our front-page story, and the link to our article is tiny. Most people just read the AI answer and move on,” said Anne Mitchell, publisher of the Methow Valley News in Washington state. “That’s ad impressions and potential subscribers we never get.”

The Defenses: Fair Use and Innovation

Legal analysts expect Microsoft and OpenAI to mount a vigorous fair use defense. The four-factor test in U.S. copyright law considers: (1) the purpose and character of the use, including whether it is commercial or nonprofit educational; (2) the nature of the copyrighted work; (3) the amount and substantiality of the portion used; and (4) the effect on the potential market for the copyrighted work. The tech companies will likely argue that training is transformative because the models learn patterns of language rather than copying expression, and that the public interest in AI innovation outweighs any harm to publishers.

Yet the complaint meticulously documents instances of near-verbatim reproduction—a factor that could undermine a fair use claim. Moreover, the fourth factor—market harm—weighs heavily in favor of the newspapers, as they can show demonstrable financial injury. Earlier this year, the Copyright Office issued guidance stating that the use of copyrighted works to train AI models may not always qualify as fair use, particularly when the output competes with the original. The LNA lawsuit likely will test that guidance in court.

What Comes Next

The case has been assigned to Judge Katherine Polk Failla, who has handled several high-profile technology and intellectual property disputes. Initial case management conferences are expected within 90 days. Given the complexity, a trial could be years away, though the plaintiffs have indicated they will seek a preliminary injunction to halt further use of their content while the case is pending. Such an injunction, if granted, would mark a seismic shift in the AI industry.

For local newspaper publishers, the lawsuit is less about a payout than about survival and control. “What we really want is a seat at the table,” said Torres. “We’re not asking for charity. We’re asking for the right to decide what happens to our life’s work. If AI wants to use our journalism, it needs to pay for it, just like any other business.”

As the legal gears begin to turn, all eyes will be on how Microsoft and OpenAI respond—and whether this coalition of small voices can force a reckoning for an entire industry. For Windows users, the outcome could reshape one of the platform’s most touted features, potentially making Copilot less useful or, in a best-case scenario, supported by a sustainable ecosystem of fairly compensated content creators.

One thing is certain: the era of unchecked scraping is under judicial scrutiny, and the future of local news—and of AI itself—may hang in the balance.