Court Demands OpenAI Hand Over 20M Anonymized ChatGPT Chats in AI Copyright Dispute

A federal judge has ordered OpenAI to turn over 20 million anonymized ChatGPT conversation logs in a major copyright lawsuit, rejecting the company’s arguments that privacy concerns should limit the disclosure.

District Judge Sidney H. Stein upheld a ruling by Magistrate Judge Ona T. Wang requiring OpenAI to produce records for news organization plaintiffs, including the New York Times Company and the Chicago Tribune Company.

The publications are among 16 copyright cases consolidated in the Southern District of New York that raise critical questions about how copyright law applies to artificial intelligence.

The discovery dispute began in July when news plaintiffs requested 120 million conversation logs.

OpenAI countered with an offer of 20 million logs representing 0.5% of its preserved data which the publications accepted.

However, in October, OpenAI refused to provide the complete anonymized sample and offered only search results that implicated plaintiffs’ works.

Judge Wang sided with the news outlets in November and denied reconsideration in December.

OpenAI appealed, claiming the magistrate judge insufficiently weighed privacy concerns and that courts should order the least burdensome discovery method.

Judge Stein rejected these arguments on Monday, stating Judge Wang properly balanced privacy concerns against the material’s relevance to the litigation.

He noted no case law requires courts to choose the least burdensome discovery option.

OpenAI primarily relied on a Second Circuit securities case in which discovery of SEC call recordings was blocked due to concerns about illegal wiretapping.

Judge Stein distinguished that precedent, explaining that the call participants had stronger privacy interests because the tapes were made secretly, as reported by Bloomberg Law.

In contrast, ChatGPT users voluntarily submitted their communications, and OpenAI’s legal ownership of the logs is undisputed.

The consolidated multidistrict litigation represents a crucial legal battle that could set precedents for dozens of similar lawsuits filed by content creators against AI companies.

The outcome may determine how AI firms can use copyrighted material to train their models.

Susman Godfrey LLP, Rothwell, Figg, Ernst & Manbeck, and Loevy & Loevy represent the news plaintiffs. OpenAI is represented by Keker, Van Nest & Peters LLP; Latham & Watkins LLP; and Morrison & Foerster LLP.

Follow us on Google News, LinkedIn, and X to Get Instant Updates and Set GBH as a Preferred Source in Google.