On August 7, 2025, the U.S. District Court for the Northern District of California ruled that AI company Anthropic is not required to provide user personal information to music publishers. This ruling may influence future court decisions on how to balance legal evidence disclosure and user protection in AI tools. (Case No.: 24-cv-03811)

District Judge Susan van Keulen (Keulen) noted at the hearing that day that the music publisher accused Anthropic of mass-copying copyrighted lyrics to train its large language model Claude, but obtaining user account names and email addresses was unnecessary to advance the litigation.

She stated: “Linking the provided conversation content to specific users lacks sufficient basis and would harm third-party privacy interests.” The key information requested regarding the lyrics “has already been disclosed, and there is currently no sufficient reason to go beyond this scope.” This ruling aligns with Judge Keulen's preliminary order issued on July 31, which denied the publishers' request for user information.

Previously, OpenAI also failed to avoid a New York court order requiring it to retain all input and output data from its AI models in another copyright lawsuit filed by news publishers. Although the content sought in the two cases differs (all input and output data vs. user account information), both OpenAI and Anthropic emphasize that privacy concerns stem from users employing chatbots for sensitive inquiries related to health, emotions, and other private matters.

In 2023, Universal Music Group and other publishers sued Anthropic. In March this year, the court dismissed claims of contributory infringement, vicarious infringement, and removal of copyright management information. However, after the publishers amended their complaint in April, they reintroduced claims of direct infringement.

Judge Keulen allowed limited searches of Anthropic employees' personal accounts linked to non-company email addresses while protecting the privacy of ordinary users. The publishers argued that this would refute Anthropic's claim that it “never intended to generate lyrics” and prove the subjective intent to infringe. Anthropic's attorneys countered that the publishers had not provided sufficient evidence to support the search of employees' private accounts.

The court ordered both parties to collaborate on a list of Anthropic executives subject to investigation and allowed the company to pre-screen employees for using personal accounts to search for lyrics. Judge Keulen emphasized the need to carefully balance privacy protection with the determination of intent to infringe, noting that “the relevant conduct must have a legal connection to Anthropic.”