Oops! OpenAI just deleted important legal data in a lawsuit from The New York Times

Sam Altman’s OpenAI has lost deleted related to a lawsuit with The New York Times and other newspapers.

Kevin Dietsch/Getty Images

The New York Times and other newspapers are in a legal battle with OpenAI over using their content.Lawyers for the newspapers are searching OpenAI’s training data as part of a discovering process.OpenAI accidentally just deleted all the lawyer’s work.

An unusual setback has happened in a lawsuit against OpenAI: the company just deleted a bunch of work by the lawyers representing its opposition.

The lawsuit was filed by a group of news organizations including The New York Times and The Chicago Tribune. It accuses OpenAI of using their articles for training and violating their copyrights.

Lawyers for the news organizations have been inspecting OpenAI’s training data under a tightly controlled discovery process involving computers without an internet connection.

According to legal filings, there was a snafu last week.

“Since November 1, 2024, the News Plantifs have spent over 150 person-hours searching OpenAI’s training data for instances of the News Plaintiff’s Asserted Works. The News Plaintiffs stored the results of their searches on two dedicated virtual machines provided by OpenAI,” said a filing reviewed by Business Insider.

“On November 14, 2024, the News Plaintiffs learned OpenAI’s engineers erased all of the News Plaintiff’s programs and search result data that was stored on one of the dedicated virtual machines.”

It went on to say that OpenAI had been able to recover some of the data, but not the file structure or file names, which lawyers said made it essentially useless.

It’s unclear how the data on the server got erased, but in a second legal filing, a lawyer for the newspapers said that they “have no reason to believe was intentional.” (Hey, who among us hasn’t accidentally deleted something important?)

Newspaper lawyers are asking the judge to have OpenAI repeat their searches so that they don’t have to redo all their (costly) work.

OpenAI did not respond to a request for comment.