USA TODAY Co. sues OpenAI over training on 19 publications

Why it matters

The lawsuit tests whether a news publisher can demand payment and restrictions when it alleges an AI company used its reporting for model training without a license, with potential consequences for both training costs and traffic to original outlets.

USA TODAY Co. and several newspapers it owns sued OpenAI in Manhattan federal court on October 8th, alleging that the AI company copied hundreds of thousands of articles and other materials to train ChatGPT without permission. The complaint seeks damages of more than $250 million and a court order blocking the alleged infringement.

The suit, filed as USA Today Co. v. OpenAI Foundation, names work from 19 publications, according to the complaint and reporting on its contents. They include USA TODAY, The Tennessean, the Detroit Free Press, The Arizona Republic, The Indianapolis Star, The Columbus Dispatch and The Milwaukee Journal Sentinel. The case is pending in the U.S. District Court for the Southern District of New York, docket No. 1:26-cv-08892.

The plaintiffs allege that OpenAI used their reporting to train the systems behind ChatGPT without a license. They also argue that ChatGPT can reproduce or summarize articles in ways that substitute for readers visiting the original publications. Those are claims in a newly filed complaint; the court has not ruled on them. Reuters reported that OpenAI did not immediately comment on the filing.

For USA TODAY Co., the dispute concerns a portfolio that extends far beyond its national newspaper. The publisher says it owns more than 200 local media organizations, and its chairman and CEO, Mike Reed, previously led GateHouse Media and New Media Investment Group through their 2019 acquisition of Gannett. The lawsuit puts local and national reporting under the same licensing question: whether public availability on the internet gives an AI developer permission to use journalism in model training.

OpenAI has argued that training models on publicly available internet material is fair use. In a January 2024 statement, the company said it viewed training as fair use and described publisher opt-outs and licensing partnerships as parts of its approach to working with news organizations. That position does not resolve the specific allegations in this case. The publishers are asking a court to decide whether OpenAI's alleged copying crossed copyright boundaries and whether the resulting products compete with the sources whose work they used.

OpenAI has also made licensing agreements with publishers, including a 2024 deal with News Corp covering current and archived content from publications such as The Wall Street Journal and the New York Post. Those agreements show that licensed access is one route for bringing journalism into AI products. They do not establish whether the material alleged in this case was covered by permission, or whether training without a license is lawful.

The suit adds to a growing set of copyright disputes over AI training. Reuters has described dozens of cases against OpenAI, Anthropic and Meta, and reported in September on arguments by authors and news outlets that statements by OpenAI and Microsoft executives could undermine the companies' copyright defenses in a separate Manhattan case. The USA TODAY complaint brings a large group of local and national publications into that fight and seeks a financial remedy as well as restrictions on the alleged conduct.

For publishers, the stakes include both control over archival reporting and the economics of how readers find it. A chatbot answer that summarizes or reproduces reporting could send fewer readers to the outlet; licensing deals offer a negotiated alternative, while litigation seeks payment and limits through the courts. The complaint asks for more than $250 million, but that figure is the plaintiffs' demand, not a court finding or an award. The case will test the allegations against OpenAI's fair-use position, while leaving the broader question of how much permission AI companies need to train on journalism unsettled.