A group of newspapers, headed by the New York Times and the New York Daily News, asked a federal court in Manhattan to impose sanctions on OpenAI, alleging the AI developer misled the court about its ability to search its systems for evidence that it used millions of their articles to train its models.
In a filing submitted on Thursday, the newspapers said OpenAI told the court it could not search its large language models for their copyrighted content. The plaintiffs assert that OpenAI concealed the fact that it had conducted such searches "even before the first News Plaintiff filed suit." The filing also alleges that OpenAI deleted, or made unsearchable, billions of ChatGPT conversations relevant to the dispute.
What the newspapers are seeking
The newspapers requested sanctions that would include payment of attorneys' fees and a judicial finding that OpenAI's chat logs demonstrate the company misused their copyrighted works. The filing frames the alleged conduct as both misleading to the court and damaging to the plaintiffs' ability to pursue discovery in the litigation.
Background of the litigation
The suit was first brought by the New York Times in 2023 and accuses OpenAI and its largest financial backer, Microsoft, of using millions of the Times' articles without permission to train the large language model that powers ChatGPT. The case is part of a broader wave of litigation from copyright holders - including authors, visual artists and music labels - who allege that AI companies such as OpenAI, Anthropic and Meta Platforms used their material to train AI systems without authorization.
Allegations of contradictory statements
The newspapers' filing points to prior statements by OpenAI that it lacked tools to search datasets and output logs for copyrighted material. The filing contrasts that claim with testimony by an OpenAI employee that the company had in fact "performed multiple searches for News Plaintiffs' content," according to the newspapers' submission.
Ian Crosby, identified in the filing as the New York Times' lead attorney, is quoted saying: "For over two years, OpenAI lied to The Times, The Daily News Plaintiffs, the public, and the court. It claimed searching ChatGPT outputs for copies of The Times' and the Daily News Plaintiffs' content was infeasible, burdensome, and invasive of users' privacy - while at the same time concealing that it had already done such searches."
Responses and context
Spokespeople for OpenAI did not immediately respond to a request for comment on the newspapers' motion, according to the filing. The complaint and the sanctions request are ongoing elements of the litigation brought by the news organizations.
The allegation that OpenAI deleted or rendered unsearchable billions of ChatGPT conversations is presented in the filing as an impediment to the plaintiffs' ability to locate and present evidence about the company's use of their copyrighted works.
The court's consideration of the sanctions motion will proceed within the existing litigation framework. The filing outlines the newspapers' position that court sanctions are warranted based on the alleged discrepancies between OpenAI's public and judicial statements and the company's internal actions described in testimony.