A group of eight major U.S. newspaper publishers has filed a federal lawsuit against Microsoft and OpenAI, accusing the tech giants of improperly reusing copyrighted news articles to train their generative artificial intelligence systems like ChatGPT and Copilot.
The lawsuit, filed Tuesday in the U.S. District Court for the Southern District of New York, alleges that Microsoft and OpenAI’s AI models have been “purloining millions of the publishers’ copyrighted articles without permission and without payment.”
The plaintiffs include publishers behind major metropolitan dailies such as the New York Daily News, Chicago Tribune, Denver Post and Orlando Sentinel. All are owned by the hedge fund Alden Global Capital.
According to the complaint, OpenAI has used datasets containing text from the publishers’ newspapers to train its GPT-2 and GPT-3 large language models that underpin ChatGPT. It claims ChatGPT can then “output near-verbatim copies of significant portions” of their articles when prompted.
The lawsuit also takes issue with Microsoft’s Copilot assistant, which is integrated into products like Windows, Bing and Office. It alleges Copilot reuses publishers’ content indexed by Bing’s search engine, often without providing links back to the original sources where ads and subscriptions could generate revenue.
“We take great care in our products and design process to support news organizations,” an OpenAI spokesperson said, adding the company aims to “discuss any concerns and provide solutions” through partnerships with publishers worldwide.
Microsoft declined to comment on the pending litigation.
The legal battle highlights rising tensions between tech companies developing generative AI tools and the publishing industry over intellectual property rights. In March, The New York Times sued OpenAI over similar allegations of copyright infringement by ChatGPT.
OpenAI has argued such instances of mirroring text are “rare failures” that it is working to address through improved training methods. The startup has also struck paid content licensing deals with some major publishers recently, including Germany’s Axel Springer and the Financial Times.
For the newspaper publishers in the new lawsuit, the case represents a push to be compensated for any use of their content by powerful AI models, rather than have it repackaged without credit or monetization opportunities.
“This is an absolutely shocking violation of intellectual property rights,” said Doug Reynolds, managing partner at Morrison & Foerster, the firm representing Alden Global. “We gave these companies notice and an opportunity to negotiate licenses for this usage, but they refused, so we had no choice but to file this action.”
With generative AI applications rapidly proliferating, the lawsuit could have major ramifications for how tech companies obtain legal access to copyrighted source material for training their AI systems moving forward.
