Nearly 400 local and regional newspapers filed a federal lawsuit against OpenAI and Microsoft on June 24, 2026, accusing the tech giants of systematically copying copyrighted journalism to train and operate ChatGPT and related AI systems. The complaint, lodged in the U.S. District Court for the Southern District of New York, marks the largest collective action by news publishers against AI companies to date and ratchets up the existential legal threat facing large-language-model (LLM) development.

The coalition behind the suit includes dailies and weeklies from every state, many of them family-owned or part of small chains. They allege that OpenAI scraped millions of articles from paywalled and public websites without permission, then used them as training data for GPT models that now power ChatGPT, Microsoft Copilot, and a growing suite of AI features woven into Windows, Edge, and the Microsoft 365 ecosystem. The newspapers seek statutory damages and a permanent injunction that could force the defendants to delete models trained on their content.

The Heart of the Allegation

At the core of the complaint is a simple claim: OpenAI and Microsoft built lucrative AI businesses on the backs of local journalism, without paying for or even asking permission to use the reporting. The plaintiffs argue that articles—ranging from investigative pieces to obituaries—were ingested into training datasets and remain embedded in the models, enabling ChatGPT and Copilot to generate responses that paraphrase, summarize, or sometimes reproduce verbatim passages from the newspapers’ archives.

“Defendants have converted the collective work product of hundreds of newsrooms into a perpetual, royalty-free content engine,” the lawsuit states, according to a copy reviewed by reporters. “Every query answered by ChatGPT with information originally reported by a local newspaper erodes the incentive to pay for that journalism.”

The filing identifies specific instances where ChatGPT allegedly generated text nearly identical to copyrighted articles, including paywalled content that should have been inaccessible without a subscription. It also names Microsoft’s Copilot, which is deeply integrated into Windows 11 and web services, as an active participant in the infringement because it relies on the same underlying model stack.

This lawsuit does not arrive in a vacuum. It builds on a wave of copyright litigation that has roiled the AI sector since late 2023. The New York Times sued OpenAI and Microsoft in December 2023 over similar claims, and a group of authors including John Grisham and George R.R. Martin filed their own class action. Investment firm Alden Global Capital, which owns dozens of U.S. newspapers, launched a separate suit in early 2024. In July 2025, the Center for Investigative Reporting joined the fray, accusing OpenAI of using its motherlode of investigative work without a license.

What sets this new action apart is both its scale and its focus on local journalism. Nearly 400 mastheads binding together represents an unprecedented collective push by publishers that have historically competed fiercely. It also zeroes in on the precarious economics of community news: many of these outlets operate with skeleton staffs and depend on subscription and advertising revenue that, they say, is being hollowed out by AI-generated substitutes.

Microsoft and OpenAI have repeatedly asserted that training on publicly available data constitutes fair use under U.S. copyright law. They have also pointed to licensing agreements reached with major publishers such as Axel Springer, the Associated Press, and Le Monde. But these deals cover only a small fraction of the news ecosystem, and critics note that none of the participating outlets were consulted before their content was used to build the foundational models.

Microsoft’s Role Under the Microscope

Microsoft’s inclusion as a defendant is particularly significant. While OpenAI built and trained the GPT family, Microsoft is the largest investor—having poured an estimated $16 billion into the company as of early 2026—and the primary commercializer of the technology through Copilot. Copilot is now a native component of Windows 11, where it assists with tasks across the OS, and it is embedded in Bing, Edge, GitHub, and the Office productivity suite.

Legal experts say the suit directly challenges Microsoft’s strategy of positioning AI as a universal assistant that can answer virtually any question. If the court rules that Copilot’s responses are derivative works, Microsoft could be forced to strip out vast amounts of training data or build expensive licensing frameworks.

“This isn’t just about OpenAI’s research sandbox anymore,” said Megan Gray, a technology law professor at Georgetown University. “It’s about a product that sits on hundreds of millions of Windows desktops and is marketed as a replacement for traditional search and research. The economic impact of that product on news publishers is impossible to ignore.”

Local Journalism in the Crosshairs

For the newspaper industry, the stakes are existential. Local papers have lost more than half their newsroom jobs in the past two decades, and many communities now exist in “news deserts” with no dedicated local reporting. The plaintiffs argue that Microsoft and OpenAI are accelerating this decline by offering AI tools that can instantly generate summaries, event listings, and even full articles that sidestep the need to visit a paper’s website or buy a subscription.

The complaint points to several examples in which ChatGPT was able to recount the details of a small-town zoning dispute, a school board election result, or a high-school sports championship after being prompted with just a topic and date—information that, the plaintiffs say, could only have been sourced from their protected articles.

“Local newspapers are the canaries in the coal mine,” said Danielle Coffey, CEO of the News/Media Alliance, a trade group that has lobbied Congress for protections against AI scraping. “If we lose them, we lose the primary source of accountability journalism for most Americans. This litigation is a last resort, but the alternative—watching our members disappear one by one—is unacceptable.”

The complaint raises claims under the Copyright Act, the Digital Millennium Copyright Act (DMCA), and state unfair competition laws. The newspapers are asking the court to:

  • Declare that OpenAI’s and Microsoft’s training and operation of LLMs using copyrighted works constitutes infringement.
  • Award statutory damages for each willful act of infringement, which could run into the billions of dollars given the number of articles at issue.
  • Issue a permanent injunction requiring the defendants to remove all copyrighted content from their training datasets and to refrain from using that content in the future.
  • Compel Microsoft and OpenAI to implement technical measures that prevent ChatGPT and Copilot from regurgitating copyrighted material.

Under the defendants’ fair use defense, courts consider four factors: the purpose and character of the use (including whether commercial), the nature of the copyrighted work, the amount and substantiality of the portion used, and the effect on the potential market for the original. The newspapers are betting that the commercial nature of ChatGPT and Copilot—plus the direct competition with their own products—will outweigh arguments that AI training is transformative.

Several judges in related cases have already signaled skepticism toward sweeping fair use claims. In the New York Times case, the court refused to dismiss the suit at an early stage, noting that “verbatim reproduction of paywalled content cannot be ignored simply because the overall model may have beneficial uses.”

Industry Reaction and Public Debate

The lawsuit has drawn immediate reactions across media and tech. Journalism associations expressed solidarity, while some large newspaper chains that already struck licensing deals remained conspicuously quiet. Newsroom union leaders applauded the move, framing it as a fight for the survival of their profession.

On social media, readers debated the merits of the case. Many voiced support for the newspapers, citing alarming experiments in which AI models produced near-exact copies of breaking-news stories. Others questioned whether decades-old articles should be protected when the same facts are often available through public records. But the core legal issue—whether ingesting copyrighted content without permission to train a profit-making AI system is fair use—remains unresolved, and the sheer volume of plaintiffs may give this case unique weight.

“Microsoft and OpenAI have had years to clean up this mess,” posted tech commentator and former Microsoft engineer David Plummer on X. “Instead they’ve relied on litigation and lobbying. When 400 newspapers join hands, Congress will pay attention even if courts take forever.”

What Happens Next

Legal observers expect a protracted battle. Microsoft and OpenAI have not yet filed an official response, but both companies are known to have formidable legal teams that specialize in intellectual property. They are likely to challenge the lawsuit on jurisdictional grounds, seek to narrow the class, and argue that many of the articles were available under robots.txt or open-access licenses.

They may also attempt to persuade the court that the relief sought—deletion of trained models—is technically impossible or disproportionately harmful to innovation. In the meantime, discovery could uncover internal documents about how training data was sourced, which might expose embarrassing practices or reinforce the plaintiffs’ claims of willful infringement.

For Windows watchers, the lawsuit carries a direct product risk. A preliminary injunction that affects Copilot’s ability to generate news summaries would immediately impact a feature Microsoft has heavily promoted as a reason to upgrade to Windows 11 and use Edge. That could rattle enterprise customers and consumers alike. However, any such remedy is months or years away, and Microsoft has the resources to appeal and negotiate.

The Bigger Picture

Beyond the courtroom, the 400-newspaper lawsuit accelerates a broader reckoning over the relationship between AI platforms and the content they depend on. Congress has held multiple hearings but passed no legislation addressing AI training-data transparency or creator compensation. The U.S. Copyright Office has issued guidance stating that works created solely by AI are ineligible for protection, but it has not opined definitively on the ingestion side.

Other countries are moving faster. The European Union’s AI Act, fully enforceable from mid-2026, requires general-purpose AI providers to make public a “sufficiently detailed summary” of training-data copyright use. Japan and Singapore have carved out broad exceptions for AI training, while the U.K. has proposed a voluntary code that would allow publishers to opt out.

In the United States, this lawsuit could become the tipping point that forces either a legislative framework or a Supreme Court ruling on what the digital age owes to those who fund original reporting. For now, nearly 400 local newspapers are betting that the law will see their side of the story.