CRBC News

OpenAI Denounces NYT Over Demand for 20 Million ChatGPT Logs — Even After Judge Ordered Production

OpenAI has publicly condemned The New York Times' request for 20 million ChatGPT user logs, calling it an "invasion of user privacy." The company’s security chief urged the court to reject the demand, even though Magistrate Judge Ona Wang ordered production on Nov. 7, noting existing protective orders and OpenAI’s de-identification efforts. OpenAI has asked the judge to reconsider, arguing the logs are irrelevant and that the company lacked an adequate chance to challenge the legal precedents relied on in the order. The case is among the most advanced copyright lawsuits facing AI firms.

OpenAI challenges The New York Times' request for user logs while seeking reconsideration of a judge's order

OpenAI publicly criticized The New York Times on Wednesday for requesting 20 million ChatGPT user logs as part of the paper’s copyright lawsuit, calling the demand an "invasion of user privacy" and urging the court to reject it. The statement, authored by OpenAI's chief information security officer Dane Stuckey and titled "Fighting the New York Times' invasion of user privacy," described the request as contrary to long-standing privacy protections and common-sense security practices.

"This demand disregards long-standing privacy protections and breaks with common-sense security practices," Stuckey wrote, urging the court to deny the Times' discovery request.

What OpenAI did not emphasize in its public message is that a federal magistrate judge has already ordered production of the logs. In a Nov. 7 order, Magistrate Judge Ona Wang found it was "appropriate" for OpenAI to produce the 20 million ChatGPT logs sought by the Times for discovery, concluding that OpenAI had not shown user privacy would be inadequately protected.

Judge Wang pointed to strict safeguards governing how attorneys and experts may review discovery, as well as OpenAI's pledged "exhaustive de-identification" of the consumer logs, and wrote that OpenAI had not explained why those protections were insufficient.

Why the Times wants the logs

The New York Times sued OpenAI and Microsoft in 2023 in Manhattan federal court, alleging the companies used Times articles to train AI models and that ChatGPT sometimes reproduced reporting that is protected by copyright. The Times says a sample of up to 20 million user logs would help it analyze how users have interacted with ChatGPT over time and whether use patterns relate to alleged copying of Times content.

Discovery safeguards and OpenAI's response

Attorneys for the Times have been subject to unusually stringent discovery controls: source-code reviews on a disconnected computer in a secure room, bans on personal electronic devices, supervised entry with government-issued ID, and other restrictions. OpenAI has also committed to a comprehensive de-identification process intended to scrub sensitive user data from the logs.

Despite the Nov. 7 ruling, OpenAI filed a motion asking Judge Wang to reconsider, arguing that producing "a massive trove of irrelevant personal user conversations" is neither supported by common sense nor the Federal Rules. OpenAI's lawyers contend the company did not have adequate opportunity to explain why the judge relied on certain precedents and that further review is warranted.

Context and implications

The Times is among several news organizations that have sued OpenAI and Microsoft; this case is one of the most advanced and far-reaching copyright suits confronting AI companies. OpenAI has publicly pressed its privacy and security arguments in the media while continuing legal efforts to limit discovery.

Disclosure: Axel Springer, the parent company of Business Insider, has a content-licensing agreement with OpenAI. Representatives for The New York Times, OpenAI, and Microsoft did not immediately respond to requests for comment.

OpenAI Denounces NYT Over Demand for 20 Million ChatGPT Logs — Even After Judge Ordered Production - CRBC News