Lawyers for the New York Times and Daily News, who are suing OpenAI for allegedly scraping their work without permission to train AI models, say OpenAI engineers accidentally deleted data that could be relevant to the lawsuit. states that it did.
Earlier this fall, OpenAI agreed to provide two virtual machines to lawyers at The Times and Daily News to help them run searches for copyrighted content in the AI training set. (A virtual machine is a software-based computer that resides within another computer's operating system and is often used for purposes such as testing, backing up data, and running apps.) The publisher's attorney said: In the letter, we mention the costs incurred by the publishers and the experts they hired. Since November 1st, I have spent over 150 hours searching for OpenAI training data.
However, according to the aforementioned letter filed late Wednesday in the U.S. District Court for the Southern District of New York, OpenAI engineers erased all publisher search data stored on one of its virtual machines on Nov. 14. .
OpenAI attempted to recover the data and was mostly successful. However, because the folder structure and file names were “irretrievably” lost, the recovered data “cannot be used to determine where the news articles Plaintiff copied were constructed.” [OpenAI’s] “Model,” the letter says.
“News plaintiffs are forced to spend countless hours of labor and computer processing time rebuilding their works from scratch,” lawyers for the Times and Daily News wrote. “News Plaintiffs learned yesterday that the recovered data was unusable and required a full week's worth of work by experts and attorneys to be redone. That's why we filed this supplemental letter today. It is.”
Plaintiffs' attorneys have made it clear that there is no reason to believe the deletion was intentional. But they said the incident highlights that OpenAI is “best positioned to search its own datasets” for potentially infringing content using its own tools. There is.
We have reached out to OpenAI for comment and will update this article if we hear back.
In this and other cases, OpenAI has argued that training models using publicly available data, including articles from the Times and the Daily News, is fair use. In other words, OpenAI licenses the samples to create models like GPT-4o that “learn” from billions of e-books, essays, and other samples to produce human-sounding text. We don't believe you need to acquire or otherwise pay for it. Even if you're making money off of those models.
That said, OpenAI has signed licensing deals with a growing number of new publishers, including The Associated Press, Business Insider owner Axel Springer, the Financial Times, People's parent company Dotdash Meredith, and News Corp. However, OpenAI refused to enter into these terms. The deal is public, but one of the content partners, Dotdash, is reportedly being paid at least $16 million annually.
OpenAI does not confirm or deny that it trained its AI systems on specific copyrighted materials without permission.