Sony and Warner Put Anthropic’s Training Library on Trial

Sony Music Publishing and Warner Chappell have sued Anthropic, chief executive Dario Amodei and co-founder Benjamin Mann in northern California. The complaint alleges that copyrighted compositions were obtained and used without permission while developing Claude. Anthropic disputes the claims and says it will defend itself. At this stage, the accusations are allegations, not findings, however loudly the filing arrives carrying its own brass band.

Anthropic Watch analysis

What the publishers allege

The publishers say the alleged copying covers tens of thousands of compositions and includes torrenting, scraping and downloading copyrighted works. Reporting from The Verge and Axios says the complaint seeks statutory damages and names individual founders alongside the company.

Music copyright is particularly layered. A single song can involve separate rights in its composition, lyrics and recording, sometimes held by different parties. That means the same underlying material can create several distinct legal questions rather than one tidy invoice.

What Anthropic says

Anthropic told Axios that it disagrees with the publishers’ claims and intends to defend itself robustly. The company had not provided a detailed response to every allegation in the complaint at the time of the initial reports.

The case follows earlier copyright disputes involving AI training data, including Anthropic’s settlement with authors and publishers. A previous settlement does not decide this complaint. It does, however, ensure the court will not be examining the question in a historical vacuum.

Why it matters

The lawsuit attacks both the legal source of training material and the scale of the alleged acquisition. If the plaintiffs prove infringement and obtain substantial statutory damages, the result could affect how AI companies document datasets, negotiate licences and assign personal responsibility for collection decisions.

The difficult policy question is not whether artists deserve protection or whether models need data. It is how training rights, provenance and compensation should work at industrial scale without pretending that either creators or technology can be wished out of the room.

Our read

Treat the filing as serious evidence of a dispute, not a verdict. The useful Watcher job is to track the complaint, Anthropic’s formal answer, motions over training and fair use, and any evidence that survives adversarial testing. Court documents beat victory laps from either side.

Discussion spark: What would a workable music-training licence need to disclose about source material, compensation and model outputs?

Anthropic Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by Anthropic.