Sony Music Publishing and Warner Chappell Music recently filed a lawsuit in a federal court in Northern California naming Anthropic, CEO Dario Amodei and co‑founder Benjamin Mann as defendants. The plaintiffs allege that Anthropic used tens of thousands of copyrighted song lyrics without authorization to train the Claude family of AI models.
Plaintiffs' allegations
According to the complaint, Anthropic allegedly "torrent[ed], copied and downloaded" copyrighted lyrics and used them in model training, after which the Claude models sometimes reproduced those lyrics in responses to user queries. The publishers also assert that Anthropic obtained copies from other sources, including legitimate lyrics services such as Musixmatch and LyricFind, and from archival datasets like those derived from Common Crawl.
The complaint further alleges that Anthropic removed or altered identifying metadata from the lyric files during processing, which the publishers say deprived songwriters and rights holders of valuable attribution and context for their works.
Damages sought and scope of the suit
The plaintiffs are seeking damages totaling in the billions of dollars. The complaint requests statutory damages of up to $150,000 per infringed work and $25,000 per work for removal or alteration of identifying information. The filing cites at least 20,000 works. In parallel, Universal Music Group and Concord Music Group have also brought claims against Anthropic.
Anthropic's response and prior litigation
An Anthropic spokesperson said the company disagrees with the publishers' assertions and intends to vigorously defend itself in court.
This is not Anthropic's first copyright dispute. In a 2025 copyright case, a court found that using copyrighted works to train models could be "fair use" in certain circumstances, but it concluded that Anthropic's central library of pirate copies was not protected in the same way. That litigation was resolved when Anthropic agreed to destroy pirate copies used or intended for training and settled the matter for $1.5 billion.
The current lawsuit focuses again on how training data were acquired and how identifying information was handled — issues that were central in the earlier case and are likely to be key battlegrounds in the forthcoming proceedings.
What comes next
The course of the litigation will depend on the parties' filings and the evidence presented. The outcome could have significant financial implications for the companies involved and may set precedent on the legality of data practices used to train large AI models.
