USA Today Sues OpenAI, Joining a Growing Copyright Fight Over AI Training Data

USA Today Co. and several of its local newspaper properties have sued OpenAI, accusing the company of copying large volumes of newsroom content to train its AI systems without permission. The publisher is seeking more than $250 million in damages, making the case another major test of how copyright law applies to generative AI and the commercial use of journalism.

The lawsuit, filed Thursday, says OpenAI used “hundreds of thousands” of articles from USA Today Co. publications and caused “real and continuing” harm to the company’s news business. The complaint adds to a widening legal front against the ChatGPT maker, which now faces a growing number of lawsuits from publishers, media companies, and other rights holders over how its models were trained.

For news organizations, the case is about more than a payout. It goes to the heart of whether AI companies can build products that compete with publishers while relying on the publishers’ reporting, headlines, archives, and local coverage as training material.

What USA Today Co. is accusing OpenAI of

USA Today Co. says OpenAI copied its journalism at scale and used it to develop products that generate billions in value. According to the complaint, the company never sought permission and instead appropriated newsroom content to improve its AI systems.

The publisher argues that the alleged copying was not incidental or limited. Rather, it says OpenAI relied on a vast amount of protected material from USA Today and its network of regional newspapers, including titles such as The Tennessean, IndyStar, The Columbus Dispatch and The Oklahoman.

The lawsuit contends that OpenAI’s business success is built on large-scale copyright infringement and that the company used journalism without authorization to create highly valuable commercial products.

That framing matters because the dispute is not only about output generated by chatbots. It is also about the upstream process that powers them: the ingestion of copyrighted material during model training.

Why this lawsuit matters now

This case arrives as courts, publishers, and policymakers are still trying to define the legal boundaries of generative AI. The core question is whether training a model on copyrighted works without a license is permissible, especially when the model can later produce summaries, answers, and text that may reduce traffic to the original sources.

USA Today Co. is seeking more than $250 million in damages, a figure that signals both the scale of the alleged copying and the seriousness of the publisher’s claims. The case could also influence how other news organizations approach licensing talks with AI companies.

For publishers, the issue is straightforward: journalism costs money to produce, and AI products can now use that journalism to attract users, subscriptions, and enterprise customers. For AI developers, the argument is usually that model training requires enormous datasets and that the technology is transformative. Courts will have to weigh those competing views.

How big is the legal fight against OpenAI?

It is getting bigger by the month. USA Today Co. is only the latest in a series of plaintiffs challenging OpenAI over alleged copyright violations tied to model training and output.

OpenAI is already facing lawsuits from The New York Times, The Intercept, Ziff Davis, CBC/Radio-Canada, Encyclopaedia Britannica, Merriam-Webster, The Seattle Times and a coalition of nearly 400 local newspapers. Some of those cases focus on training data, while others also examine how AI systems respond to prompts that reference copyrighted text.

Who else is suing OpenAI?

A range of media companies and publishers have already taken OpenAI to court, spanning national newsrooms, local newspapers, and reference publishers. The breadth of the plaintiffs shows that concerns about AI training are no longer isolated to one segment of the media industry.

• The New York Times
• The Intercept
• Ziff Davis, the owner of CNET
• CBC/Radio-Canada
• Encyclopaedia Britannica
• Merriam-Webster
• The Seattle Times
• Nearly 400 local newspapers in a coalition case

Each lawsuit may differ in detail, but the larger pattern is the same: media companies are arguing that generative AI developers built commercially successful systems by exploiting protected content without negotiation or compensation.

What does the complaint say about OpenAI’s business?

USA Today Co. argues that OpenAI used its articles to create products worth hundreds of billions of dollars. The complaint says that the company did so without asking for permission and without paying for a license.

That claim is aimed at the economics of AI, not just the mechanics of training. Publishers say the value of their archives lies in the reporting labor, editorial judgment, and verification that went into the stories. If AI systems can ingest that value at scale without compensation, the publishers argue, the incentives for original reporting could weaken over time.

The lawsuit also suggests that the harm is ongoing, not historical. Even if the original copying happened during model development, publishers say the downstream effects continue as AI tools draw attention and revenue away from traditional news products.

How the case fits into the broader copyright debate

This dispute is part of a larger collision between copyright law and machine learning. AI companies need immense amounts of text to train large language models, but much of the most useful text is copyrighted and commercially valuable.

That tension has already reached courts in multiple countries. In the United States, judges are being asked to decide whether model training resembles a protected transformative use or whether it amounts to unauthorized copying that harms the market for original works.

For publishers, the concern is not abstract. Newsrooms spend heavily on reporting, editing, fact-checking, and distribution. If AI systems can use that content to answer user questions directly, publishers may lose both audience attention and revenue.

For AI firms, the defense often rests on the idea that training does not simply reproduce works, but instead extracts patterns from them to create new tools. That distinction will likely be central in many of the pending cases.

Timeline of the OpenAI publisher dispute

The legal conflict between OpenAI and the publishing industry has unfolded quickly. The sequence below shows how the pressure has built.

Why local news organizations are especially concerned

Local outlets may have the most to lose if generative AI systems absorb their work without compensation. Unlike national brands that may have broader revenue streams, regional newspapers rely heavily on subscription income, local advertising, and the audience value of original reporting.

The lawsuit includes a number of local titles, underscoring that this is not just a battle over big-city media brands. It is also about whether regional journalism can survive in a market where AI tools can summarize, repurpose, or potentially compete with local reporting.

Local reporting is expensive because it is labor-intensive. City hall coverage, court reporting, sports coverage, and neighborhood news are all costly to produce. If those stories become raw material for AI training without payment, local publishers argue the incentive to invest in that work could shrink further.

What OpenAI has said so far

OpenAI did not immediately respond to a request for comment on the USA Today lawsuit. That means the company’s formal legal response is not yet public in this case.

In previous copyright disputes, OpenAI has generally argued that its systems are lawful and that AI training falls within permissible uses of data. The company has also pointed to partnerships and licensing deals in some contexts, suggesting that it sees licensing as part of the future ecosystem even if it disputes liability in court.

For now, the absence of an immediate response leaves the complaint as one-sided. The company will likely answer in court filings, where it can challenge the claims, the damages request, and the legal theory behind the case.

What could happen next?

The case is likely to move through a familiar sequence: complaint, response, motions to dismiss, and possibly lengthy discovery over what data was used to train the models and how that material was obtained. That process could take months or years.

Several outcomes are possible:

1. The parties could reach a licensing settlement.
2. A judge could narrow or dismiss some claims before trial.
3. The case could proceed to a broader ruling on training-data copyright issues.
4. The dispute could encourage more publishers to negotiate paid access agreements with AI companies.

Whatever the outcome, the lawsuit adds to the pressure on AI developers to prove that their business models are legally durable, not just technically impressive.

How this could reshape AI licensing deals

One likely effect of cases like this is more bargaining power for publishers. Even if they do not win every lawsuit, media companies may use the threat of litigation to push for licensing revenue, attribution controls, and restrictions on how their content is used.

That could lead to a market in which large AI firms pay for premium news archives, current reporting, or real-time access to articles. Some deals of that kind are already emerging across the industry, although the terms are often confidential.

At the same time, the litigation pressure may force AI companies to diversify their data strategies. That could mean more licensed content, more synthetic data, or more emphasis on public-domain and open-web material.

Why the outcome will matter beyond one publisher

The legal precedent from this case could reach well beyond USA Today Co. and OpenAI. If courts decide that training on copyrighted journalism without consent is unlawful, the decision could reshape the AI sector’s access to high-quality text data. If courts rule more broadly in OpenAI’s favor, publishers may have a harder time forcing compensation through litigation.

Either way, the case sits at the center of a bigger economic question: who gets paid when artificial intelligence is trained on the work of professional writers, editors, and newsrooms?

For now, USA Today Co. has made its position clear. It believes OpenAI used its reporting to build a highly profitable product, and it wants a court to hold the company accountable.

Key facts at a glance

The legal battle over AI and copyright is no longer theoretical. With another major newspaper publisher now in court, the industry’s reckoning over how generative systems are trained is becoming broader, more expensive, and harder to avoid.