The music industry’s dispute with generative AI has expanded from music-making platforms to the data behind a general-purpose model.
Sony Music Publishing, Warner Chappell Music and affiliated publishers filed a copyright complaint in the Northern District of California on August 28 against Anthropic, CEO Dario Amodei and co-founder Benjamin Mann.
The complaint alleges that Anthropic obtained copyrighted works through torrenting, scraping and downloading, then used those works to develop and operate its Claude models. Axios reports that the case concerns tens of thousands of musical compositions, making its stated scope broader than earlier publisher claims against the company.
The complaint is an allegation, not a judgment. Anthropic says it disagrees with the claims and will defend itself in court.
Why a composition case matters
A recording can contain several distinct rights. The composition covers the song itself—its music and lyrics—while the master covers a particular recorded performance. Ownership can be divided among writers, publishers, artists and labels.
That layered system is why the same underlying song can create more than one dispute about training. It is also why this case reaches beyond the question of whether an AI can generate audio that resembles a performer. The publishers are challenging how protected musical works were allegedly acquired and copied to build the model.
No court has resolved those claims. For artists and listeners, the immediate result is not a new creative rule but another high-stakes test of whether AI companies need permission—and on what terms—before copyrighted music enters a training pipeline.