A federal court in Manhattan has ordered OpenAI to submit millions of anonymized ChatGPT user logs in the ongoing major copyright case involving The New York Times and other media organizations.

U.S. Magistrate Judge Ona Wang stated that nearly 20 million chat logs are directly relevant to the publishers’ claims, and turning them over will not endanger user privacy. She rejected OpenAI’s argument that sharing the logs could violate confidentiality.

Wang also noted that multiple layers of security are already in place due to the sensitive nature of the data.

An OpenAI spokesperson referred to an earlier statement from the company’s Chief Information Security Officer, Dane Stuckey, who argued that the Times’ request “overlooks established privacy protections” and goes against basic security practices.

OpenAI has already appealed the order to U.S. District Judge Sidney Stein.

The New York Times did not immediately comment on the ruling. Meanwhile, Frank Pine, executive editor of MediaNews Group — also part of the lawsuit — said that OpenAI was “hallucinating” if it believed it could hide evidence showing its business model relies on “stealing from hardworking journalists.”

The lawsuit, filed by the Times in 2023, is among several cases accusing OpenAI, Microsoft, and Meta of using copyrighted material without permission to train AI models. Publishers argue that the logs are essential to determine whether ChatGPT reproduced their copyrighted content and to refute OpenAI’s claim that they “hacked” the chatbot.

OpenAI maintains that releasing the logs could expose confidential user information and that “99.99%” of the chats have no connection to the case.

Judge Wang reaffirmed that OpenAI’s de-identification process and other security measures are sufficient. She ordered the company to provide the logs within seven days after removing user-identifying information.