OpenAI slams court order that lets NYT read 20 million complete user chats

December 3, 2025

TL;DR

OpenAI is challenging a court order to produce 20 million user chats to The New York Times in a copyright infringement lawsuit.
OpenAI argues the disclosure of complete conversations is too broad and risks exposing private user information.
The company claims over 99.99% of the chats are irrelevant to the case.
The New York Times states the lawsuit is about accountability for using copyrighted works and that user privacy is safeguarded by anonymization and a protective order.
ChatGPT conversations have previously been found in Google search results, raising broader privacy concerns.
OpenAI plans to implement advanced security features, including client-side encryption, to protect user data.
A judge previously ordered the production of the de-identified chats, citing existing protective orders and anonymization measures.
OpenAI claims the court did not acknowledge its arguments regarding the limitations of the de-identification process.
The New York Times disputes OpenAI's claims, stating OpenAI's terms of service allow training on user chats and turning them over for litigation.
OpenAI suggested alternative methods for the NYT to find relevant chats, which were rejected.
The NYT argues that immediate production of output logs is essential for their analysis of how the product functions and user interactions.
OpenAI contrasts the current demand with earlier requests that were limited to chats related to 'Times content'.

Continue reading
the original article