Six local-news publishers sue Microsoft and OpenAI over alleged scraping of journalism for AI training

What happened

On Oct. 2, 2026, six small publishing companies filed a copyright complaint in the U.S. District Court for the Southern District of Mississippi (No. 3:26-cv-00791) against Microsoft and ten OpenAI entities. The plaintiffs are Emmerich Newspapers, Ojai Media, Coopwood Publishing Group, Coopwood Magazine, Coopwood Media Group and Coopwood Newspapers; Emmerich is described as one of the largest privately owned newspaper chains in Mississippi, with papers in Louisiana and Arkansas as well.

What the complaint alleges

The publishers allege the defendants crawled their sites and copied tens of thousands of articles -- including content behind paywalls -- without permission or payment, repeatedly as models were updated. The complaint cites OpenAI's own training-data descriptions (WebText for GPT-2; Common Crawl, WebText2, Books1, Books2 and Wikipedia for GPT-3) and alleges analyses of Common Crawl and OpenWebText found the plaintiffs' material. It further alleges that author credits, publication names, copyright notices and terms of use were stripped from pages before training, pointing to the Dragnet and Newspaper content extractors OpenAI described using for WebText.

The complaint's three counts do not apply to every defendant the same way: Count I (direct infringement, 17 U.S.C. §501) targets Microsoft and the OpenAI entities; Count II (vicarious infringement) targets Microsoft and five OpenAI entities, tying Microsoft to storing and processing training data on its supercomputing platform and distributing output through Copilot; Count III (removal of copyright management information, 17 U.S.C. §1202(b)) is aimed at the OpenAI entities, not Microsoft.

The plaintiffs seek statutory damages (including for alleged willful infringement), compensatory damages, restitution, disgorgement, costs and attorneys' fees, and a jury trial. The most unusual request is an injunction under 17 U.S.C. §503(b) ordering removal of all copies of the registered works from all GPT or other LLM models and training sets.

These are allegations in a complaint; nothing has been decided. No response from Microsoft or OpenAI has been located.

Context

This is another local-news entry in the wave of publisher suits over AI training data, filed in a different federal district (Mississippi) than the New York consolidated proceedings and brought by a small group of independent publishers. Whether it stays in Mississippi or is consolidated or transferred is not known.