OpenAI asks judge to reverse order to hand over 20M ChatGPT conversations to NYT publishers

What happened

OpenAI has asked a federal judge in New York to reverse a court order requiring it to hand over roughly 20 million anonymized ChatGPT conversations to The New York Times and other publisher plaintiffs as part of the consolidated copyright-infringement litigation in the Southern District of New York.

Details

• In the filing, OpenAI argued that producing the chat logs would expose confidential user information, claiming 99.99% of the transcripts have no connection to the copyright claims.
• OpenAI's chief information security officer Dane Stuckey condemned the order as "an invasion of user privacy" in a blog post, saying it would force turnover of tens of millions of personal conversations from people with no connection to the case. OpenAI said it had previously resisted broader demands for 1.4 billion chat logs and had proposed privacy-preserving alternatives (targeted searches for chats containing NYT text), which the publisher rejected.
• The NYT and other outlets say the data is crucial to determine whether ChatGPT reproduced their copyrighted material and to rebut OpenAI's claim that they "hacked" the chatbot to manufacture evidence.
• Magistrate Judge Ona Wang, who previously ruled in favor of disclosure, has said the chat data would undergo exhaustive de-identification and other safeguards. The order covers a random sample of consumer ChatGPT conversations from December 2022 to November 2024; enterprise, education, business and API customers are not affected.
• A NYT spokesperson countered that privacy remains protected since the court ordered only anonymized chat samples under a protective order.

Status

The original disclosure order was covered as a separate item (2026-10-05). This filing is a new procedural development challenging it.