USA Today Joins the Latest Publisher in Legal Action Against OpenAI

Meta description: USA Today Co. and local papers sue OpenAI, alleging hundreds of thousands of articles trained AI models and seeking more than $250 million in damages.

Tags: OpenAI, USA Today Co., AI copyright, publisher lawsuits, AI training data

Featured image: Abstract technology illustration with layered document fragments and glowing network nodes, no real people, no company logos, no text.

Background: A growing line of publisher claims against OpenAI

USA Today Co., together with several local newspapers it owns, is suing OpenAI in a copyright lawsuit that centers on the company’s AI models and the text used to train them. The filing alleges that OpenAI copied “hundreds of thousands” of articles, and it asks for damages of more than $250 million. The claim connects a familiar publishing asset, newsroom text, with a newer question for courts and the technology industry: whether large language models, AI systems trained on massive amounts of text, can be built from protected writing without a license.

The case arrives after a string of copyright lawsuits filed against OpenAI. The New York Times has already sued, and OpenAI is also facing legal action from The Intercept, CNET owner Ziff Davis, CBC/Radio-Canada and Encyclopaedia. Those earlier actions show that the dispute is not limited to one national newspaper or one business model. It spans local newsrooms, digital publishers, public media, reference content and other forms of editorial writing.

Development: Thursday filing sharpens the money claim

The filing was reported earlier by Reuters and covered by The Verge. It was submitted on Thursday, and its core demand is a specific monetary one. USA Today Co. and its local newspapers say the unauthorized use of their content “has done real and continuing” harm to their outlets.

The number “hundreds of thousands” matters because it frames the alleged dataset not as a few sampled headlines, but as a large corpus of articles. The $250 million figure matters because it moves the case from a general principle dispute into a damages claim. If the case proceeds, the parties will likely need to discuss how article counts, licensing rates, publication reach, market harm and model use are measured.

For readers, the practical effect is that AI training data is becoming a commercial asset with a price. Publishers have traditionally sold access through subscriptions, syndication and licensing agreements. AI developers have built models on broad text corpora. The lawsuits test whether those two models can coexist without a new licensing layer.

Latest progress: The case sits inside a broader litigation wave

USA Today Co. is now the latest plaintiff to put OpenAI’s training practices under legal pressure. The filing does not, by itself, decide whether copying articles for AI training infringes copyright or qualifies as fair use, a doctrine that allows limited unlicensed use of copyrighted material in certain cases. It sets out allegations and a damages request. The next stages typically involve responses, motions and evidence about how articles were collected, stored, transformed into training data and reflected in outputs.

The timeline also shows how quickly the legal front has expanded. Earlier lawsuits created a pattern; the Thursday filing adds a major publisher group with local newspapers and a concrete ask for more than $250 million. That combination gives the case a clearer economic profile than a purely principle-focused claim.

What to watch next

The key moments will be OpenAI’s response, any court rulings on whether the claims can proceed, and the emergence of licensing or settlement discussions. For newsrooms, the outcome could affect whether publishers can negotiate paid access for model training. For AI developers, it could affect how they document datasets and seek permissions. For readers, the result may shape whether AI answers remain built on freely scraped text, licensed news archives, or a mix of both. At present, the USA Today Co. filing remains an active claim within a broader set of OpenAI copyright lawsuits, not a final judgment.