News orgs want OpenAI to dig up millions of deleted ChatGPT logs

In a significant legal development, OpenAI is facing pressure from news organizations to disclose 20 million ChatGPT logs, a move that could have far-reaching implications in the ongoing copyright dispute. The U.S. District Judge Sidney Stein has denied OpenAI's objections, which argued that the privacy of users not involved in the litigation should take precedence over the needs of the plaintiffs. Judge Stein upheld the previous ruling by Magistrate Judge Ona Wang, emphasizing that while privacy concerns were important, the need for transparency in the case outweighed them. OpenAI had proposed a less invasive method of producing logs, suggesting that it could filter the data using search terms to limit access only to relevant conversations. However, the judge found that the proposed approach did not adequately address the plaintiffs' needs. Stein pointed out that the number of logs to be produced had already been significantly reduced from billions to 20 million, and assured that all identifying details had been removed from the shared chats. The judge acknowledged that even logs that do not directly reproduce the works of the news organizations could still be pertinent to OpenAI’s defense based on fair use. Moreover, OpenAI's argument for a less intrusive method was dismissed as it failed to provide adequate legal support for its claims. Judge Stein affirmed that the order to produce the full sample was justified and did not represent a legal error, thus paving the way for the plaintiffs to gain access to critical information in their case against the AI company.