Overview
- The publisher coalition, which filed a complaint on June 24, 2026, represents nearly 400 local and regional outlets and seeks statutory damages, injunctions, and other relief in the Southern District of New York.
- The suits allege defendants used automated crawlers and extraction tools named Dragnet and Newspaper to remove copyright management information and copy article text into training datasets such as WebText, WebText2, and C4.
- The New York Times filed an amended complaint that adds a contributory-infringement theory, accusing Microsoft of building a bespoke supercomputing system to enable large-scale ingestion of Times material for model training.
- OpenAI and Microsoft say their models are trained on publicly available data and invoke fair use, while Google is separately pressing publishers to join News AI pilots that may grant broad content rights in exchange for fees.
- If courts or deals require paid licenses or tighter limits on data use, smaller newsrooms could lose referral traffic and subscription revenue and AI companies would face higher training costs and tighter product design constraints.