USA Today Co. and 13 affiliates sue OpenAI, seeking more than $250m

USA Today Co. and 13 affiliated publishing companies sued OpenAI in federal court in Manhattan on Thursday 8 October 2026, alleging that OpenAI trained its models on their journalism without permission and repackages it for ChatGPT users. The complaint, filed in the Southern District of New York as case 1:26-cv-08892, says the plaintiffs "seek damages in excess of $250 million". None of the allegations has been tested in court.

Who is suing whom

The caption lists 14 plaintiffs: USA Today Co., Inc. and 13 companies the complaint says are "all owned by USA TODAY Co., Inc.", including Gannett Satellite Information Network, Detroit Free Press, Inc., Journal Sentinel Inc. and Phoenix Newspapers, Inc. Together, the complaint says, they own copyrights in content from "19 different publications", among them USA Today, The Arizona Republic, the Detroit Free Press, IndyStar, The Columbus Dispatch and the Milwaukee Journal Sentinel.

The defendants, as captioned, are OpenAI Foundation; OpenAI GP, LLC; OAI International, Inc.; OpenAI OpCo, LLC; OpenAI Global, LLC; OAI Corporation; and OpenAI Group PBC.

The complaint brings three counts: copyright infringement, vicarious copyright infringement, and removal of copyright management information under the Digital Millennium Copyright Act. The plaintiffs ask for a jury trial.

What the complaint alleges

• Training. It alleges that OpenAI's models "were trained on copyrighted material scraped from the internet without authorization—regardless of paywalls or other access restrictions". "Upon information and belief", it says, the models "have copied hundreds of thousands of articles and other materials" from the plaintiffs' publications.
• Datasets. It says the plaintiffs' content makes up "more than 160,000 entries" in WebText, an OpenAI dataset used for GPT-2, and "over 122 million tokens" in C4, which it describes as "a filtered English-language subset of a 2019 snapshot of Common Crawl".
• Outputs. In what it calls "illustrative and non-exhaustive examples", set out in full in Exhibit B, the complaint describes prompts in which GPT-5.6 retrieved articles from papers including the Indianapolis Star and the Detroit Free Press and produced long summaries. In one case, it says, the output was "an extensive multi-section summary paraphrasing and following the same structural organization as the original article".
• Traffic. It quotes an OpenAI software engineer as writing that "no matter how prominently we show the links, users won't click", citing a filing in the consolidated OpenAI copyright litigation.
• Wilfulness. It argues that OpenAI's licensing deals with other news organisations, which it says include the Associated Press, Axel Springer, The Atlantic and Vox Media, show that OpenAI "understands copyrighted journalism has value and that using it requires permission and payment".

The complaint cites a statutory ceiling of "up to $150,000 for each willful copyright infringement, plus up to $25,000 per violation" for stripping copyright management information. Among the relief it requests is an order for "destruction under 17 U.S.C. § 503(b) of all GPT or other LLM models and training sets that incorporate the USA TODAY Plaintiffs' content", as well as a permanent injunction.

OpenAI's side

OpenAI's response was not found in the sources we read. Trademark Lawyer Magazine reported that "OpenAI did not immediately respond to Reuters' request for comment on the complaint." The same report says OpenAI and other technology companies "have argued in related litigation that using copyrighted works to train AI models can constitute fair use under US copyright law". We did not find a USA Today Co. press release; the company's website blocked our reader.

The complaint itself places the case alongside "the long list of copyright holders who have filed lawsuits against OpenAI and other AI companies", many of them consolidated in the same court. As we reported in September, the US Justice Department filed a statement of interest in that consolidated case arguing that training large language models on copyrighted text should be treated as fair use.

The UK connection

USA Today Co. owns Newsquest, which describes itself as "the leading local news publisher in the UK with a portfolio of more than 200 brands". Newsquest's own website says that "In July 1999, Newsquest was acquired by USA TODAY Co. (formerly Gannett)". Newsquest is not among the 14 plaintiffs named in the complaint, and none of the 19 listed publications is a UK title.

Why it matters

On our reading, the suit adds a 19-title US newspaper group to the publishers taking OpenAI to court, and it targets the outputs of a current model, GPT-5.6, as well as past training. The complaint's request that models be destroyed is a remedy the plaintiffs are asking for, not one any court has granted. Whether training and summarising of this kind can be fair use, as OpenAI and others have argued in related litigation according to Trademark Lawyer Magazine, or is infringement, as the publishers allege, is for the court to decide.