USA Today Co., the parent of USA Today (formerly Gannett), sued OpenAI in federal court in the Southern District of New York on October 8, alleging that it used hundreds of thousands of news articles from its 19 newspapers without authorization to train its GPT series of models, and seeking more than $250 million in damages. It is another important piece in the copyright war between publishers and AI companies.
Key points
1. Detailed evidence, with the complaint aimed squarely at training data: The 79-page complaint says OpenAI's internal WebText corpus contains more than 160,000 article entries from the group, 83,000 of them from usatoday.com alone, and that content from its publications' domains in the C4 dataset exceeds 122 million tokens. The complaint also cites internal OpenAI communications: co-founder Greg Brockman said the model was "especially good at predicting news article text," and a research vice president said "training networks is making them memorize data."
2. GPT-5.6 output samples as exhibits: The complaint attaches examples of GPT-5.6 output showing that, when asked to find articles by headline, the model can produce long multi-section summaries that reproduce the structure and organization of the originals, involving articles from local papers such as The Indianapolis Star and the Detroit Free Press. The plaintiff argues that OpenAI deliberately trained the model to be a "substitute" for news, as the head of ChatGPT admitted: "once ChatGPT gives the answer, there's no reason to click the original link."
3. Joined to the consolidated Southern District copyright litigation: The same day, the group filed a notice of related case asking that the suit be joined to the multidistrict litigation (MDL) over OpenAI's copyright infringement already consolidated in that court, whose plaintiffs already include The New York Times, the Center for Investigative Reporting and other organizations. The outcome could reshape the industry's rules on "web scraping plus model training."
Source: read the original. This text is a summary for reference only.
