According to Music Business Worldwide, Reservoir filed a lawsuit against Anthropic on October 21, 2026, in the U.S. District Court for the Northern District of California. Reservoir alleges that Anthropic copied lyrics from 853 compositions to train Claude without permission and that Claude can generate copies or derivative works based on those lyrics. The reported damages demand totals $151,407,500, not a court award.

What Reservoir alleges about Claude

Reservoir alleges that lyrics from 853 compositions were copied into training data for Claude and that the AI model can reproduce or create derivative works based on those lyrics in response to prompts. Those claims are allegations in the lawsuit, not findings of infringement.

The reported complaint names Joni Mitchell’s “Big Yellow Taxi” and “Both Sides Now,” Madonna’s “Papa Don’t Preach,” Billy Strayhorn’s “Lush Life,” and Evanescence’s “Bring Me to Life.”

How the reported damages demand adds up

The reported total combines a statutory-damages figure with two claims under the Digital Millennium Copyright Act (DMCA), a U.S. law that includes provisions concerning access controls and copyright-management information.

Reported damages componentAmountWhat the claim concerns
Statutory damages$127,950,000Reported amount for 853 works; the complaint seeks up to $150,000 per work alleged to have been willfully infringed.
DMCA claim$2,132,500Alleged circumvention of access controls.
DMCA claim$21,325,000Alleged removal of copyright-management information.
Total of reported amounts$151,407,500The sum of the three reported figures, sought in the complaint.

The arithmetic is $127,950,000 + $2,132,500 + $21,325,000 = $151,407,500. Reservoir also seeks actual damages and Anthropic’s profits, or statutory damages instead. The amounts are requests in the complaint, not damages awarded by a court.

Other relief Reservoir seeks

Reservoir also seeks an injunction and the impoundment or destruction of unauthorized copies. The requested relief includes copies allegedly retained in training datasets, internal repositories, model weights, and server infrastructure.