Music Production
Sony Music Publishing and Warner Chappell have filed a lawsuit against AI developer Anthropic, alleging illegal use of copyrighted music to train AI models, seeking damages up to £115,000 per work Photo by Andreas Forsberg on Unsplash

Anthropic is facing a federal copyright lawsuit in California from major music publishers Sony Music Publishing and Warner Chappell, which accuse the artificial intelligence company of illegally copying songs by artists including The Beatles, Taylor Swift and Michael Jackson to train its Claude models.

The publishers are seeking statutory damages of up to $150,000 for each copyrighted work they say was infringed.

According to legal filings submitted on Friday in the US District Court for the Northern District of California, the publishers are seeking statutory damages of up to $150,000 for each copyrighted work they allege was infringed.

The legal action follows disputes across the creative industries over how generative platforms source data, with the publishers alleging that Anthropic conducted a 'brazen campaign' of mass digital piracy.

The lawsuit names Anthropic co-founders Dario Amodei and Benjamin Mann alongside the company, claiming leadership actively directed the harvesting of protected lyrics and sheet music to train their artificial intelligence software.

Massive Torrenting Operations Alleged

In court records, the publishers claim Anthropic gathered thousands of protected musical compositions through illicit digital repositories. The filing asserts that the company acquired copyrighted materials by systematically scouring unauthorised databases, specifically naming Library Genesis and Pirate Library Mirror.

Through these operations, Anthropic allegedly torrented and downloaded catalogues containing sheet music and lyrics for tracks including Bon Jovi's 'Livin' On A Prayer', Earth, Wind & Fire's 'September', Jerry Lee Lewis' 'Great Balls Of Fire', The Allman Brothers' 'Ramblin' Man' and Leonard Cohen's 'Hallelujah'.

The lawsuit contends that this unlicensed data ingestion enables Claude to generate identical or near-identical lyrics when prompted by users.

Whether reproducing songs such as Marvin Gaye's 'Ain't No Mountain High Enough', Survivor's 'Eye Of The Tiger', Mariah Carey's 'All I Want For Christmas Is You' or Swift's 'Paper Rings', the model is alleged to act as an unauthorised substitute for the original works.

The publishers argue that training the system on copyrighted material enables Claude to produce lyrics that compete directly with professional songwriters.

Escalating Legal Battles

Anthropic moved to reject the accusations. In an official statement, the company said it disagreed with the publishers' claims and intended to defend itself in court.

The case comes after a June 2025 ruling in which a federal judge held that Anthropic's use of copyrighted books to train Claude constituted fair use, but separately found that the company's copying and storage of more than seven million pirated books infringed copyright.

The filing also cites Anthropic's earlier £1.15 billion ($1.5 billion) agreement in September 2025 to resolve a class-action lawsuit brought by book authors.

Sony and Warner argue that the startup treats such penalties as a cost of doing business.

They say a settlement of that scale cannot deter unlawful conduct by a company that, according to the publishers' complaint, has reached a 'staggering 2-trillion-dollar valuation'.

The legal challenge follows litigation filed by Universal Music Group, which sued Anthropic in 2023 and again earlier this year over similar lyric-related claims.

Damages Bid Could Reach Hundreds of Millions

Requesting a jury trial, the music publishers say that new technologies must be developed within existing copyright law.

By pursuing statutory damages of up to £115,000 ($150,000) for each infringed work, the potential liabilities facing Anthropic could reach hundreds of millions of pounds if a court finds wilful infringement.

The complaint adds to a growing set of disputes between generative artificial intelligence developers and rights holders, following lawsuits against companies such as OpenAI by The New York Times and Encyclopaedia Britannica.

As federal judges examine how machine-learning models use creative archives, the California courts will consider whether scraping unlicensed musical scores and lyrics falls within lawful use or constitutes copyright infringement.