Sony and Warner Music file new AI training lawsuit against Anthropic, alleging copyrighted songs were illegally used to build Claude

Sony Music and Warner Music have filed a new lawsuit accusing AI developer Anthropic of training its Claude model on copyrighted music without authorization, according to a report by Billboard published Monday.

The complaint alleges that Claude was “built on copyright theft,” with the music companies contending that training data included songs and other protected content they say were not licensed for AI use. The filing focuses on what the companies describe as unauthorized harvesting of music catalog assets to fuel the model’s capabilities.

Billboard’s report says the dispute includes allegations tied to well-known holiday repertoire, as well as songs associated with major contemporary artists. The companies’ claims, as characterized in the report, span examples ranging from iconic Christmas songs to Taylor Swift hits.

The filing adds to a series of high-profile lawsuits and demands for clarity on what copyright law permits for machine-learning training, particularly when output systems can generate music-adjacent content. It also underscores the growing scrutiny on how large AI labs acquire and process content at scale.

For Sony and Warner, the case is framed around enforcement of rights and control over the use of their catalogs, with the companies seeking legal remedies for what they characterize as unauthorized appropriation of protected material. The complaint’s central contention is that Anthropic used copyrighted works in training, not merely in customer-facing access or licensed consumption.

Anthropic has not been described in the report as agreeing with the allegations. As with other AI copyright disputes, the outcome will likely depend on questions that courts address in these cases, including how training data is collected, whether specific copyrighted inputs were used, and what role (if any) licensing and fair-use arguments play under the facts presented.

The lawsuit also arrives as record labels and rights holders are simultaneously negotiating licensing frameworks for AI-related products, while AI developers face increasing pressure to document data provenance. If the case proceeds, the next steps would typically include service of the complaint, responses by Anthropic, and early procedural rulings that can shape how much discovery is allowed and which claims move forward.