<\/p>\n
Anthropic is pushing a fair use argument to dismiss a lawsuit from music publishers over its use of song lyrics in training Claude, signalling a potential shift in AI copyright disputes amid industry tensions.<\/p>\n<\/div>\n
Anthropic is seeking to shut down a copyright case brought by music publishers by arguing that its use of song lyrics to train Claude falls within fair use, pressing a legal theory that has already gained traction in a separate authors\u2019 dispute. In a redacted summary judgment filing reported by Digital Music News, the company said training on protected text is transformative and consistent with copyright\u2019s broader purpose of encouraging new expression rather than freezing existing works in place.<\/p>\n
The motion leans heavily on Anthropic\u2019s earlier courtroom success in June 2025, when a federal judge found that training large language models on copyrighted books could qualify as fair use because the process produces something meaningfully different from the source material. According to Digital Music News, Anthropic is now trying to extend that reasoning to lyrics, while also pointing to the dismissal of a separate Meta case that nevertheless stopped short of blessing AI training as lawful across the board.<\/p>\n
That background matters because the music publishers\u2019 case has become one of the more closely watched tests of how far AI firms can push the fair use argument. Digital Music News reported that Anthropic\u2019s authors settlement in August 2025 resolved the book-focused class action before trial, but left the publishers\u2019 suit alive, with a hearing on summary judgment due on 15 July 2026 and a trial later scheduled for 28 September 2026.<\/p>\n
Anthropic is also trying to blunt the publishers\u2019 damage claims by arguing that the disputed lyrics are widely available online and that Claude is designed to resist verbatim reproduction. In the company\u2019s telling, most of the prompt-and-output examples gathered in discovery came from the publishers or their agents, who allegedly tried to bypass the system\u2019s safeguards. The filing also says the publishers have dropped contributory and vicarious infringement claims after recent case law narrowed the path for holding service providers liable for user conduct.<\/p>\n
The dispute is part of a wider legal fight over whether training data can be licensed, or whether AI developers can rely on fair use instead. Digital Music News noted that Universal Music Group and Sony Music are now seeking discovery in their separate case against Suno after Warner Music struck a licensing deal with the AI music generator in November 2025, undermining Suno\u2019s claim that no real market exists for such agreements. Anthropic is facing still more pressure from the industry as well: in March 2026, BMG filed a separate complaint accusing the company of mass infringement, piracy and removal of copyright management information.<\/p>\n
Inspired by headline at:<\/strong> [1]<\/a><\/sup><\/p>\n