News publishers seek sanctions against OpenAI in copyright battle

The New York Times, the New York Daily News and other publishers asked a Manhattan federal judge on Thursday to sanction OpenAI, accusing the company of withholding evidence in a copyright fight over how ChatGPT was trained on news content.

The news organizations told the court that OpenAI failed to preserve or disclose datasets and ChatGPT logs that could show exactly how the company used their articles during training. A recent deposition by an OpenAI employee conflicted with earlier company statements about whether those materials could be searched. The publishers accuse OpenAI of lying to the court about its ability to search its systems for proof that it misused millions of articles in AI training.

The case is being heard in the Southern District of New York before U.S. Magistrate Judge Ona T. Wang, who previously ordered OpenAI to preserve ChatGPT output logs. Later rulings required the company to produce large samples of de-identified chat logs, despite OpenAI’s privacy objections. OpenAI argues that indefinite retention of consumer ChatGPT and API data would conflict with its privacy commitments to users.

The fight began with The New York Times’ lawsuit in late 2023. Since then, the litigation has expanded to include the Daily News, MediaNews Group, the Chicago Tribune, Ziff Davis and the Center for Investigative Reporting. OpenAI says training on digitized books, articles and other online materials is protected by fair use.

The dispute has also become a fight over the economics of digital publishing. The publishers argue that AI chatbots are competing with news outlets as a source of information while reducing the web traffic that funds journalism, a pressure that has grown since Google added AI-generated summaries to search results.