Music publishers challenge Anthropic over song lyrics used to train Claude AI chatbot

Music Publishers Intensify Legal Battles Against AI Companies

Major music publishers have escalated their Anthropic lawsuit, asking a California judge to reject the AI startup's potential fair use defense and rule that copying song lyrics to train AI models violates copyright law. Universal Music Group, Concord, and ABKCO filed a motion on Monday in federal court in San Jose, arguing that the AI-powered chatbot Claude unlawfully reproduces their lyrics on demand without permission1

1

. The publishers contend that Claude's AI-generated lyrics are derivatives that "compete with and dilute the market" for their original works, making this unauthorized use of copyrighted material a clear case of AI copyright infringement3

3

.

Separately, music rights management company BMG filed its own lawsuit against the Amazon and Google-backed AI company on March 18, alleging that Anthropic copied lyrics from hit songs by the Rolling Stones, Bruno Mars, Justin Bieber, and Ariana Grande to train the large language models powering Claude2

2

. BMG cited 493 examples of allegedly infringed copyrights and is seeking damages of up to $150,000 per work4

4

.

The Fair Use Doctrine Takes Center Stage

The central question in these legal battles against AI companies revolves around whether AI fair use applies when training AI models on millions of copyrighted works. Music publishers argue their case differs significantly from previous rulings. U.S. District Judge William Alsup previously ruled in a separate case that Anthropic's use of books for AI training was "quintessentially transformative," siding with the company1

1

. However, the music publishers told U.S. District Judge Eumi Lee that their record of the Claude chatbot reproducing their work on demand is "overwhelming," distinguishing their claims from those involving books3

3

.

The publishers have asked the court to rule before trial that suing AI startup Anthropic is justified because the company infringed their copyrights when it illegally used copyrighted lyrics. BMG claims that Anthropic used automated scraping tools to copy text from public websites and illegal online pirate libraries since the company's founding in 2021 by former OpenAI staffers4

4

. "Anthropic's practice of training AI models on copyrighted works sourced from unauthorized torrent sites, among other acts, stands in direct opposition to the standards required of any responsible participant in the AI community," BMG stated2

2

.

Mounting Pressure on AI Companies

These lawsuits represent just a fraction of dozens of disputes between copyright owners—including authors and news outlets—and tech giants like OpenAI, Microsoft, and Meta Platforms over training data for their AI systems1

1

. Anthropic was the first major AI company to settle one such case, agreeing to pay a group of authors $1.5 billion last year to resolve a class-action lawsuit covering approximately 500,000 works, paying roughly $3,000 per book5

5

.

The original music publishers' lawsuit against Anthropic was filed in 2023, alleging infringement of copyrights in lyrics from at least 500 songs by musicians including Beyoncé, the Rolling Stones, and the Beach Boys1

1

. Universal Music Group later expanded its claims, now seeking statutory damages exceeding $3 billion for the infringement of more than 20,000 songs5

5

.

BMG's complaint demonstrates how Claude readily supplied users with complete or substantial portions of chart-topping songs like "Uptown Funk" by Bruno Mars, "You Can't Always Get What You Want" by the Rolling Stones, and Louis Armstrong's "What A Wonderful World"4

4

. The lawsuit notes that BMG sent a cease-and-desist letter in December 2025, but Anthropic did not respond or engage in licensing discussions4

4

.

Implications for AI Development

The outcomes of these cases will likely shape how AI companies approach intellectual property law and whether they can continue relying on the fair use doctrine when building their systems. BMG argues in its complaint that "Generations of inventors have brought revolutionary new products to market while complying with copyright law. Anthropic's rapid development of its new technology is no excuse for its egregious law-breaking"5

5

. The company is also seeking a court order requiring Anthropic to disclose details about its training data, methods, and model capabilities, including identifying any BMG-controlled material used5

5

.

Divergent rulings from judges in Northern California suggest courts are still grappling with how to apply existing copyright frameworks to AI technologies. As these cases progress through the legal system, they will test whether AI companies can transform copyrighted material into something new or whether they must obtain licenses before using protected works for training purposes. The music industry's aggressive stance signals that rights holders are unwilling to accept unauthorized derivatives of their catalogs, regardless of the technological innovation involved.