USA Today Co. sues OpenAI for $250m and seeks model destruction

A US newspaper group is asking a court to do more than make OpenAI pay. USA Today Co. wants the models themselves destroyed. The company and 13 affiliated companies, which publish 19 newspapers, sued seven OpenAI entities in the Southern District of New York, seeking more than $250 million, according to Law Commentary.

The papers include The Arizona Republic, Detroit Free Press, The Tennessean, Milwaukee Journal Sentinel and Indianapolis Star. The 79-page complaint is dated 8 October 2026 and was reported on 12 October. It alleges OpenAI copied hundreds of thousands of news articles without permission to train its models, and that ChatGPT reproduces their reporting. Its examples use GPT-5.6.

The complaint puts numbers on the alleged copying. It cites more than 160,000 entries from the papers' sites in OpenAI's WebText dataset, about 83,300 of them tied to USA Today's website. It also brings a DMCA claim, alleging OpenAI stripped author names, titles and other ownership information from the articles.

The relief list is broad: more than $250 million, an injunction, profits attributable to the copying, attorneys' fees and a jury trial. The most consequential item is the request, under Section 503(b) of the Copyright Act, for destruction of GPT and other models and training datasets containing the publishers' material.

On fair use, the publishers argue that ChatGPT's commercial use and its potential to substitute for news content undermine the defence. They say that substitution could cut traffic and subscription, advertising and licensing revenue. OpenAI has defended similar suits on fair-use grounds. The New York Times sued OpenAI and Microsoft in December 2023 over similar claims. Law Commentary's report did not include an OpenAI response, and it did not link the filing.

The plaintiffs have asked the court to treat the case as related to the consolidated OpenAI copyright litigation.

The damages figure is a number OpenAI can model. The destruction request is not. By asking for GPT models and training datasets to be destroyed, the publishers move the dispute from a price for past use to a threat against the product line, and that shifts bargaining power in any licensing talks.

The revenue argument is also aimed at the commercial heart of the case. The publishers frame ChatGPT as a substitute that erodes subscriptions, advertising and licensing income, which is the argument they need against fair use. Other news publishers weighing their own claims will watch how the court treats it. Watch whether the court accepts the related-case request and whether OpenAI's answer contests the model-destruction remedy directly.