case-law

The News/Media Alliance Suit: Publishers Coordinate Against OpenAI, Microsoft, and Cohere

By Nora Whitfield · August 9, 2026

On June 24, 2026, a coalition of publishers organized by the News/Media Alliance filed suit against OpenAI and Microsoft. If that sounds familiar, like a repeat of the New York Times case, look at the plaintiff list before you assume that. Advance Local Media, Condé Nast, The Atlantic, Forbes Media, The Guardian, Business Insider, the Los Angeles Times, Politico, and Vox Media are all named. That's not one institutional plaintiff with deep pockets and a dedicated legal team. That's a trade association coordinating a lawsuit on behalf of a wide, deliberately diverse cross-section of American journalism, local and national, general-interest and business-focused, legacy print and digital-native.

Why the size and shape of the plaintiff group matters

The New York Times suit against OpenAI and Microsoft has functioned, fairly or not, as the flagship case people point to when they talk about publishers versus AI. It's easy to read a single, high-profile plaintiff as an outlier, a uniquely resourced newspaper with a uniquely strong claim taking a swing that smaller publishers couldn't afford to take themselves. This suit removes that framing. When Condé Nast, The Guardian, Politico, and the LA Times are all named plaintiffs in the same complaint, organized through the same trade association, it stops looking like one outlier and starts looking like the industry's actual position: essentially every kind of publisher with content valuable enough to be worth scraping believes it has the same claim.

The News/Media Alliance itself represents a large swath of the newspaper industry, local papers as much as national ones, and its involvement here signals this isn't limited to the handful of publications famous enough to get their own headlines when they sue. It's coordinated, industry-wide litigation, which changes the settlement calculus for OpenAI and Microsoft considerably. A single plaintiff can settle quietly. An entire trade coalition settling sets a public benchmark every other publisher will then expect.

The DMCA claim that sets this suit apart

The complaint alleges the standard training-data copyright infringement claims you'd expect, unauthorized use of published journalism to train GPT models. But it also includes a specific Digital Millennium Copyright Act claim: that OpenAI stripped copyright management information, author bylines, copyright notices, terms-of-use language, from the content before feeding it into training. That's a distinct legal theory from "you used my article without permission." It's closer to "you used my article without permission, and you also removed the information that would have told anyone downstream that permission was required and who to ask for it."

This is the same category of claim, DMCA 1202(b), central to the Doe v. GitHub Copilot litigation over open-source code, now showing up in the publisher context. Courts across multiple industries are being asked essentially the same underlying question: when an AI system trains on protected content and the identifying and licensing metadata attached to that content disappears somewhere in the process, does that stripping itself create liability separate from whatever happens with the content afterward. A win on this theory here would matter well beyond this one case, since it's a claim that doesn't depend on proving the AI's output reproduces your specific text, only that your attribution and rights information was removed along the way.

The Cohere suit: the litigation wave reaches past the famous names

There's a related development here too: some of the same publishers involved in this suit, including Advance Local and Condé Nast, along with The Atlantic and others, have also sued Cohere Inc. Cohere isn't a household name the way OpenAI, Anthropic, Google, or Meta are. It's an enterprise-focused AI company building models for business customers rather than consumer chatbots, and it draws far less press attention. The complaint against Cohere alleges what press coverage has described as "widespread unauthorised use of publisher content in developing and running its generative AI systems," essentially the same core allegation as the OpenAI suit, aimed at a company most consumers have never heard of.

That detail matters more than it might seem. It means publishers aren't only targeting the companies famous enough to be household names. They're pursuing the same claims against lower-profile AI vendors building similar products with, apparently, similar sourcing practices. If your AI company trains on scraped web content and isn't OpenAI or Anthropic, this suit is a signal that obscurity isn't a defense. The News/Media Alliance and its member publishers appear to be treating this as a pattern to litigate against wherever they find it, not a grudge against one or two famous defendants.

What this means for AI authors

If you're a publisher or a writer whose work has appeared anywhere with a byline, credit line, or copyright notice attached, this suit is worth watching for what it says about the scale of coordinated response now forming in your industry. A trade association organizing dozens of publishers into one complaint, alongside a separate, parallel suit against a completely different company using the same theory, tells you this isn't a handful of media companies testing a novel legal theory anymore. It's becoming the industry's standard playbook.

If you're building a product on top of any AI model, whether you built the model yourself or you're a customer of one, the Cohere suit specifically is the reminder that "we're not one of the famous AI labs" doesn't put you outside the blast radius of this kind of claim. The same sourcing questions that expose OpenAI expose any company using similar training data practices, regardless of how well known the company is.

And if you're on the content-creation side, wondering whether your published work might already be sitting in someone's training set, the DMCA angle in this suit is a useful thing to understand: even if proving direct copyright infringement from training is hard, proving that your byline and copyright notice were stripped out along the way is a narrower, more concrete claim, and it's exactly the kind of specific, factual question that a real publication record, showing what you wrote, when, and under what terms, helps answer.