USA Today sues OpenAI for $250M over 'hundreds of thousands' of copied articles

USA Today Co. filed suit against OpenAI on Thursday, accusing the AI developer of training its models on "hundreds of thousands" of the publisher's articles without authorization. The complaint, filed alongside several of the local newspapers USA Today owns, asks for damages exceeding $250 million. Reuters first reported the filing.

The suit adds one of the country's largest newspaper chains to a growing list of copyright plaintiffs targeting OpenAI. The New York Times, Chicago Tribune, New York Daily News and a group of authors have filed similar cases, arguing that the company's training pipeline relies on newsroom work it never licensed or paid for. USA Today's complaint bundles its national newsroom with regional titles, a structure that could widen the pool of allegedly infringed material and the potential damages.

At the center of the dispute is how OpenAI assembled its training corpora. Publishers contend that scraped web content, including their own archives, became model weights without consent, and that the resulting chatbots can reproduce passages or compete directly with the outlets that produced the reporting. OpenAI has maintained that training on publicly available text is fair use, a defense that has yet to be settled in court.

The $250 million figure gives the case a concrete price tag at a moment when several publishers have chosen licensing over litigation. OpenAI has struck content deals with News Corp, The Associated Press, Axel Springer and others, paying for access to archives while fighting the holdouts in court. USA Today's decision to sue rather than sign suggests the licensing terms on offer did not match what the chain believed its archive was worth.

For the AI industry, the case reinforces that training data provenance is now a legal liability, not just an engineering detail. Every new model release invites fresh scrutiny of what went into the dataset, and companies that cannot show clean sourcing face accumulating exposure across multiple jurisdictions. The outcome could shape how much AI developers pay for news archives — and whether future systems are built on licensed corpora or contested ones.

What to watch next is whether OpenAI moves to settle, as it has with some media partners, or litigates the fair-use question toward a ruling that would apply industry-wide. A consolidated loss or win on training data could reset the economics for every developer that scrapes the open web. Court scheduling and any motion to dismiss will signal how quickly that reckoning arrives.