OpenAI accidentally removes potential evidence in the training data process

In a stunning mistake, OpenAI engineers accidentally deleted crucial evidence collected by Fresh York Times and other major newspapers in their lawsuit over artificial intelligence training data, according to a court filing Wednesday.

The newspapers’ legal teams spent more than 150 hours combing through OpenAI’s artificial intelligence training data to find cases where their news stories were included, the filing claims. However, it does not explain how this error occurred or exactly what data it contained. While the filing indicated that OpenAI acknowledged the error and attempted to recover the data, what was recovered was incomplete and unreliable – so what was recovered cannot aid properly trace how the news organizations’ articles were used to building OpenAI artificial intelligence models. While OpenAI lawyers characterized the data deletion was deemed an “error” – lawyers for The Fresh York Times noted that they had “no reason to believe” it was intentional.

The Fresh York Times Company launched this landmark battle last December, alleging that OpenAI and its partner Microsoft built their artificial intelligence tools by “copying and exploiting millions” of articles in the publication, and now, as a result, “directly compete” with its content . The publication seeks to hold OpenAI liable for “billions of dollars in statutory and actual damages” for allegedly copying its works.

The Times has already spent more than $1 million fighting OpenAI in court, a significant sum that few publishers can afford. Meanwhile, OpenAI has struck deals with major retail outlets such as Axel Springer, Conde Nast and The Verge parent company Vox Media, suggesting that many publishers prefer to cooperate rather than fight.

OpenAI declined to join The Fresh York Times in filing an update with the court. The statement was made by Jennifer Maisel, a lawyer representing the news organizations, to formally notify the court of what had happened.

In an email to EdgeOpenAI spokesman Jason Deutrom said the company disagrees with the characterizations and will provide its own response soon. The Fresh York Times refused Edgerequest for comment.