Elvis Presley's music owners sue Anthropic for allegedly using 274 songs to train its models, seeking $41M+; Anthropic says it disagrees and will defend itself
The $41 million lawsuit contends that the A.I. company used Presley's music to train its models without permission.
Context & Ripple Effects
The Presley claim adds a catalog-specific case to Anthropic's longer dispute with music publishers, which began with a 2023 lawsuit by UMG, Concord and ABKCO and expanded in January 2026 when a publisher coalition sought $3 billion over more than 20,000 songs. A California federal judge had rejected a preliminary effort to stop Anthropic's lyrics training in 2025, leaving the underlying copyright questions unresolved.
By August 2026, Sony Music and Warner Chappell's lawsuit alleged training on tens of thousands of songs. The new action narrows that broader conflict to a highly identifiable repertoire while adding a separate damages claim; Anthropic disputes the allegations and says it will defend itself.
First-order effects
- Elvis Presley's rightsholders open a separate claim seeking more than $41 million over the alleged use of 274 songs, while Anthropic must defend its training-data practices in another music-copyright proceeding.
- The case puts the alleged use of a defined Presley catalog at issue, rather than only the larger song datasets asserted in the publishers' earlier complaints.
Second-order effects
- The additional Presley action strengthens music rightsholders' use of catalog-by-catalog litigation alongside industry-wide publisher suits, increasing the number of claims Anthropic must address independently.
- For music publishers and AI developers, the dispute makes rights clearance and evidence about training inputs more central to commercial negotiations over generative-music and language-model uses.
Third-order effects
- If courts continue to entertain claims tied to identifiable repertoires, copyright ownership becomes a practical control point over which cultural data AI developers can use and on what terms.
- The pattern favors licensed generative supply as a route to reduce training-data exposure, though the pending cases have not established the governing legal standard.
The trend: Music rights holders are using both broad publisher actions and repertoire-specific claims to press AI developers toward clearer training-data permissions.