A coalition of publishers representing nearly 400 local and regional newspapers filed a sweeping federal lawsuit on June 24, 2026, against OpenAI and Microsoft, alleging that the technology giants illicitly used copyrighted journalism to train Copilot and other AI models. The complaint was lodged in the U.S. District Court for the Southern District of New York, setting the stage for one of the largest collective copyright actions yet brought by the news industry against artificial intelligence developers.

The suit, brought by publishers who collectively own properties ranging from small-town weeklies to major metro dailies, charges that Microsoft and OpenAI scraped, reproduced, and exploited millions of articles without seeking permission, offering compensation, or even acknowledging the source material. It marks a dramatic escalation in the legal offensive that content creators have mounted against generative AI systems, and it arrives as local journalism struggles with cratering ad revenues and newsroom layoffs.

The Heart of the Allegations

According to the complaint—obtained by windowsnews.ai from court records—the publishers claim that OpenAI’s GPT models, and by extension Microsoft’s Copilot, were trained on vast corpora that included verbatim or closely paraphrased copies of their stories. They assert that this unauthorized ingestion violates the Copyright Act and constitutes massive, willful infringement.

The plaintiffs are not a single large media conglomerate but an alliance of companies that own small and mid-sized papers. Together, they oversee roughly 400 titles spread across dozens of states. By banding together, they hope to pool resources and create a united front that can match the legal firepower of two of the world’s most valuable corporations.

The complaint zeroes in on the “text and data mining” process used to assemble training datasets. It alleges that OpenAI and Microsoft knowingly utilized content from paywalled sites, breaking through access controls or simply ignoring the terms of use that prohibit automated scraping. Furthermore, the publishers argue that Copilot—deeply integrated into Windows, Edge, and Microsoft 365—directly competes with news outlets by sometimes regurgitating large portions of articles in response to user queries, eroding the need to visit publishers’ own websites.

The lawsuit echoes earlier, high-profile complaints. In late 2023, The New York Times sued the pair, alleging “billions of dollars in statutory and actual damages.” That case, still unresolved, challenged the entire premise that training AI on copyrighted material falls under the “fair use” doctrine. Other publishers, including the owners of the Daily News and Chicago Tribune, have also filed suit. What sets the new action apart is its sheer scope: hundreds of newspapers with a cumulative readership in the tens of millions.

Legal experts have been closely watching these suits. The core question is whether scraping publicly available (but copyrighted) text to train AI constitutes fair use—a transformative purpose that adds new meaning—or rather a straightforward infringement that requires licensing. The U.S. Supreme Court’s 2023 decision in Andy Warhol Foundation v. Goldsmith, which narrowed fair use in the context of visual art, may lend weight to publishers’ arguments. And while some content creators have opted to negotiate licensing deals (e.g., The Associated Press and Axel Springer), many smaller outlets feel they have no bargaining power.

The Crucial Role of Local Journalism

Beyond the legal theories, the complaint paints a deeply pragmatic picture. Local newspapers, it states, invest thousands of dollars in reporting that holds local officials accountable, covers school boards, and chronicles community life. When AI systems hoover up that work and regurgitate it without any link or payment, they siphon away the very traffic and subscription revenue that keep those newsrooms alive.

Over the past two decades, more than 2,500 local newspapers have ceased publication in the United States. The survivors operate with skeleton staff. The publishers argue that if AI companies are allowed to freely exploit their content, it will accelerate the “news desert” crisis, leaving communities without reliable, original reporting.

“This is not about opposing innovation,” one plaintiff’s executive told windowsnews.ai on condition of anonymity because the litigation is pending. “We simply cannot allow two of the richest companies in history to take what we create without paying a fair price. That is a highway to the destruction of local journalism.”

Microsoft’s Copilot: The Window to the World

Microsoft’s Copilot, built on OpenAI’s technology, is a particular focus of the suit. Copilot is woven into Windows 11, the Edge browser, and Office apps, making it the AI assistant for hundreds of millions of people. When a user asks Copilot to summarize the day’s events, it may deliver a neatly packaged digest drawn from copyrighted articles, without ever directing the user to the original publisher’s site.

The complaint cites specific examples: a user querying for “latest city council vote on zoning” received a detailed, paragraph-long answer that closely tracked an exclusive story published by a local paper. The AI response did not mention the paper’s name or provide a link. The publishers contend that Microsoft designed Copilot precisely to keep users inside its own ecosystem, thereby boosting engagement and advertising revenue for Microsoft while stealing audience from news sites.

For its part, Microsoft has long maintained that it respects copyright and has implemented tools to help publishers control how their content is accessed. The company has argued that training AI on publicly available text is lawful and analogous to a person reading a news article and learning from it. OpenAI, too, has said that it believes in fair use and that it is working with news organizations to find mutually beneficial models.

A Timeline of Tensions

The lawsuit did not materialize out of nowhere. Tensions between publishers and AI firms have simmered since 2023. After the debut of ChatGPT, many news sites began blocking OpenAI’s crawler, GPTBot, via their robots.txt files. But the plaintiffs in this suit allege that by the time they implemented such blocks, the damage was done; their earlier archives had already been ingested.

Some outlets tried negotiation before litigation. A number of local newspaper groups approached Microsoft and OpenAI over the past year, seeking licensing agreements similar to those struck by larger media companies. Those talks, the complaint suggests, went nowhere. Instead, the tech companies offered vague proposals for “partnerships” that would provide technology tools but no cash compensation for past or ongoing use of news content.

Frustration boiled over earlier this year when a coalition of small publishers sent a joint letter to the CEOs of OpenAI and Microsoft, urging them to enter formal copyright licensing negotiations. When no substantive response arrived, the coalition voted to file suit.

What the Plaintiffs Want

The complaint asks the court to award statutory damages for each infringed work, which under U.S. law can run up to $150,000 per work in cases of willful infringement. With potentially hundreds of thousands of articles at issue, the total exposure for the defendants could climb into the billions of dollars.

Beyond monetary relief, the publishers seek a permanent injunction that would bar OpenAI and Microsoft from using their content without a license in the future. They also demand the destruction of any AI models or training sets that incorporate their copyrighted material—a remedy that, if ever ordered, would have seismic consequences for the entire AI industry.

The Industry Reacts

News of the lawsuit rippled through media and tech circles. The News/Media Alliance, a trade group representing thousands of publishers worldwide, issued a statement praising the plaintiffs for “standing up for the fundamental principle that creators must be compensated for their work.” The group renewed its call for federal legislation that would compel AI companies to pay for training data.

On the other side, technology industry associations warned that such lawsuits threaten innovation. The Computer & Communications Industry Association, of which Microsoft is a member, argued that using publicly available data to train AI is “longstanding, globally accepted, and essential for the development of trustworthy AI.” Some legal scholars have cautioned that an overly broad reading of copyright could stall progress not only for large corporations but also for open-source AI projects.

Open source advocates point out that if training on copyrighted text requires a license, only deep-pocketed firms could afford to build foundation models, further concentrating power. This tension between protecting creators and ensuring a competitive AI landscape is at the heart of the global policy debate.

The case, styled as “[Lead Plaintiff] et al. v. OpenAI, Inc. et al.,” will be assigned to a federal judge in the coming days. Both sides are expected to file preliminary motions, with Microsoft and OpenAI likely to move for dismissal or, failing that, to argue that the claims should be narrowed. Discovery, if it proceeds, could unearth internal documents revealing how the companies built their training sets and whether they deliberately circumvented paywalls or copyright notices.

Legal observers say the suit’s outcome could hinge on whether the court accepts the fair use defense. Under the four-factor test, a judge would weigh the purpose and character of the use, the nature of the copyrighted work, the amount and substantiality of the portion used, and the effect of the use on the potential market for the original. The last factor—market harm—may be the publishers’ strongest card, as they can point to declining traffic and subscriptions that they attribute to AI summaries.

Even if the case settles—as many believe it eventually will—the pressure is mounting on Congress to act. Several bills have been introduced to create a new compulsory licensing scheme for AI training, somewhat like the system that governs music royalties. The publishers’ suit could provide fresh momentum for such legislative efforts.

The Bigger Picture: What This Means for Windows Users

For the millions of people who use Windows and Microsoft’s AI tools daily, the lawsuit underscores an uncomfortable tension. Copilot is marketed as a productivity booster, a generator of creative content, and a reliable research assistant. Yet the information it provides is only as good—and as legal—as its sources.

If courts rule that Copilot’s outputs infringe copyright, Microsoft might be forced to alter how its AI generates responses. Features like the “Deep Search” in Bing, which already weaves together information from multiple sources, could be restricted, potentially making the web less useful for consumers. Conversely, a ruling that allows the practice could embolden further AI development but at a cost to the very news organizations that feed the AI with fresh, factual content.

Windows enthusiasts have a stake in this fight. Many rely on Copilot for news summaries, code generation, and document drafting. If the tool becomes less capable or if publishers pull their content from AI training pools en masse, the utility of AI in Windows could diminish. The integration of AI into the operating system is a cornerstone of Microsoft’s strategy; anything that disrupts that could affect future updates and features.

What the Plaintiffs Must Prove

To succeed, the publishers must first establish that they own valid copyrights in the articles. That is likely straightforward in most instances. They must then show that OpenAI and Microsoft actually copied their material—a task made easier if the outputs contain identifiable strings or paraphrases that closely match the originals. The complaint hints at having conducted systematic tests, feeding AI prompts designed to retrieve specific articles, and documenting verbatim or near-verbatim regurgitations.

The harder challenge is overcoming fair use. The publishers will argue that AI training is not transformative because the output serves the same basic function as the input: providing information to readers. The fact that Copilot can summarize or answer questions does not, in their view, add sufficiently new expression or purpose; it merely repackages existing journalism.

Microsoft and OpenAI will counter that their models learn statistical patterns from the data and do not store or replay articles in their entirety. They will point to the public benefit of AI and insist that any short excerpts that do appear in outputs are de minimis.

The Cost to Local Newsrooms

Lost in the legal arguments is the human toll. The publishers’ complaint includes declarations from editors and reporters whose jobs have been cut even as their stories continue to be read—by AI. “It is surreal to see your byline, your words, being served up by a machine that has never called a source, never sat through a three-hour city council meeting, never knocked on a door,” one veteran reporter said.

The suit asks the court to see news articles not as mere data points but as the product of significant investment and labor. If the law cannot protect that investment, the plaintiffs warn, the supply of original reporting will dry up, and AI systems will eventually be left scraping their own synthetic content—a phenomenon researchers call “model collapse.”

International Implications

The U.S. case is being watched overseas. In Canada, Australia, and the European Union, governments have already moved to require tech platforms to pay news publishers for content. The new lawsuit could spur similar claims in other jurisdictions or lead to stricter regulation of AI training globally.

Microsoft, being a U.S. company with a dominant position in enterprise AI, faces unique exposure. The outcome of the New York case could force a reckoning for the entire industry, pushing more companies to negotiate license agreements before training rather than after. Some analysts predict that regardless of the court’s ruling, the major cloud providers will soon offer “copyright-safe” AI services that only use licensed or fully open data, much as they now offer HIPAA-compliant healthcare clouds.

Conclusion: A Defining Moment

As the lawsuit works its way through the federal courts, it will test the boundaries of copyright law in the age of artificial intelligence. For the owners of nearly 400 newspapers, it is a fight for survival—an attempt to claim a seat at the table before the digital transformation sweeps them away entirely.

For Microsoft and OpenAI, the stakes are almost as high. A loss could mean billions in damages and a fundamental reengineering of their AI architectures. A win, conversely, could reinforce the status quo and accelerate the deployment of AI across every sector.

For users of Windows and Copilot, the case is a reminder that the technology we take for granted rests on a legal and ethical foundation that is itself still under construction. The resolution, when it comes, will shape not only how we read the news but how our machines learn to understand the world.