• by Wendy Davis , April 30, 2026

Anthropic's use of lyrics to train its large language models is a "fair use" that doesn't infringe on music publishers' copyright, tech industry organizations argue in court papers filed this week.

"Training a generative AI model is a paradigmatic application of that principle that is already being deployed in a wide range of innovative contexts and applications," say groups including NetChoice and the Computer & Communications Industry Association in a friend-of-the-court brief filed with U.S. District Court Judge Eumi Lee in the Northern District of California.

"Developers engage in intermediate, non-expressive copying of works -- not to enjoy, display, or distribute them, but to build a fundamentally new kind of tool: one that analyzes statistical relationships among words and word fragments to enable human users to create new expression," the groups add.

They are weighing in on a closely watched lawsuit filed in October 2023, when music publishers including Universal Music Group, Concord Music Group and ABKCO alleged that Anthropic's “mass copying and ingestion” of song lyrics infringed copyright.

“Anthropic has built its business by unlawfully taking and using massive amounts of copyrighted content without permission or credit -- including publishers’ lyrics,” the music companies alleged.

The publishers added that the Anthropic's Claude chatbot generated “identical or nearly identical” copies of lyrics of songs including “What a Wonderful World,” “Gimme Shelter,” and “American Pie,” in response to prompts by users.

The music companies recently urged Lee to rule that Anthropic did not make fair use of the lyrics, arguing that it trained Claude to serve users with copies of lyrics in response to prompts.

Anthropic countered that its use of the songs is "transformative" and protected by fair use principles.

"Musicians write lyrics for artistic expression. Anthropic uses that same material to help Claude understand human language and enable progress and productivity in science, business, and education," the company argued in an April 20 motion seeking summary judgment -- meaning ruling in its favor before trial.

Anthropic also said it has "guardrails" aimed at preventing users from reproducing copyrighted materials, and that it strengthened those guardrails after learning of the publishers' allegations.

NetChoice and other tech industry groups contend in their friend-of-the-court brief that training generative artificial intelligence "is among the most clearly transformative uses a court has been asked to evaluate."

The organizations write: "The copyrighted works are not copied to be consumed or distributed; they are analyzed to extract statistical patterns that enable a tool to generate new text or other output at the direction of a human user. Both the purpose and the character of this use are categorically different from the original purpose and character of the works."

Last June, a federal judge in the Northern District of California partially sided with Anthropic in a separate lawsuit over its use of copyrighted material to train Claude. He ruled that Anthropic didn't infringe copyright by digitizing books it had purchased, and then using them for training purposes, writing that doing so "was exceedingly transformative and was a fair use."

But Alsup also ruled against Anthropic regarding claims that it downloaded millions of pirated books, ruling that those downloads were not protected by fair use. Anthropic later agreed to settle those claims for $1.5 billion.