Anthropic reaches settlement with music publishers over lyrics dispute

Anthropic has entered into an agreement to settle part of a copyright infringement lawsuit brought against model creator Claude AI for allegedly distributing protected song lyrics. The contract has been signed was issued Thursday by U.S. District Judge Eumi Lee, ordering Anthropic to apply existing guardrails to training future artificial intelligence models and establish a procedure for music publishers to intervene in cases of suspected copyright infringement.

In October 2023, several music publishers, including Universal Music Group, ABKCO, Concord Music Group, and Greg Nelson Music, filed a copyright infringement lawsuit against Anthropic for allegedly training its artificial intelligence system on lyrics from at least 500 protected songs. According to the complaintwhen Claude was asked to provide lyrics to songs such as Beyoncé’s “Halo,” Mark Ronson’s “Uptown Funk” and Maroon 5’s “Moves like Jagger,” the chatbot provided responses “containing all or a significant portion of these lyrics.”

While music publishers acknowledged that platforms such as Genius already distribute song lyrics online, they noted that these sites pay a licensing fee for the operate of protected songs. According to the lawsuit, Anthropic “intentionally removed or altered copyright management information” from the affected works while processing data found on these sites to train its artificial intelligence models.

Under the agreement signed by Anthropic on Thursday, the artificial intelligence company says it will maintain guardrails already in place aimed at preventing its artificial intelligence models from infringing copyrighted content. Anthropic will also apply its existing barriers to any future AI systems it develops. The music publishers and Anthropic will work together in “good faith” to resolve any cases where guardrails are found to be ineffective, and the court will stand ready to resolve any disputes.

Claude “is not intended to be used for copyright infringement and we have multiple processes in place to prevent such infringement,” Anthropic said in a statement to Hollywood reporter. “Our decision to join this resolution is consistent with these priorities. We continue to look forward to demonstrating that under current copyright law, using potentially copyrighted material to train generative AI models is the quintessential fair use.”

The music publishers who filed the original complaint have filed a motion for a preliminary injunction prohibiting Anthropic from training future models based on the lyrics of their protected songs, and the court is expected to rule in the coming months.