Sony Music Publishing and Warner Chappell Sue Anthropic Over AI Training Data

Introduction

The copyright battle surrounding generative AI is expanding beyond books and news into the music industry. Sony Music Publishing and Warner Chappell have filed a complaint against Anthropic in the US District Court for the Northern District of California. The publishers say that tens of thousands of copyrighted works were used in the development, training, or operation of Claude. Their description of the conduct is part of the complaint and remains subject to the court’s review.

Key points

• The publishers are asking for up to $150,000 per work. They are also seeking up to $25,000 for each instance in which identifiable copyright-management information was allegedly stripped.
• If the court accepted the maximum figures across the works at issue, total damages could reach several billion dollars.
• The complaint names Anthropic as well as co-founders Dario Amodei and Benjamin Mann as individual defendants.
• According to the filing, Mann used BitTorrent to download more than five million pirated books, while Anthropic employees downloaded at least two million additional books from Pirate Library Mirror. These are allegations made by the plaintiffs, not established findings.
• The publishers also allege that Anthropic scraped lyrics from sites including MusixMatch and LyricFind, which license content from music rights holders.
• Songs identified in the complaint include “Ain’t No Mountain High Enough,” “Livin’ On a Prayer,” “September,” “Hallelujah,” and Taylor Swift’s “Paper Rings.”

Why it matters

The dispute is about more than whether particular songs can be reproduced by a model. It raises questions about how training material was acquired, whether works were copied during data preparation, and whether copyright information was retained or removed. Those questions are especially complicated for music, where a single track can involve songwriters, music publishers, recording companies, collecting societies, and licensed digital services.

For Anthropic, the complaint adds to a growing list of copyright disputes. The company recently settled a case brought by the publishing industry for $1.5 billion and has also faced suits involving Universal Music Group, Concord, ABKCO, BMG, and Round Hill Music. The accumulation of cases could push AI companies to build more auditable datasets, preserve rights metadata, and negotiate broader licensing arrangements before training or serving models.

The public account currently comes primarily from the publishers’ filing. Anthropic had not immediately responded to a request for comment in the source report. The court will ultimately need to assess which works were copied or used, who carried out the alleged downloads and scraping, whether the conduct is protected by a legal defense such as fair use, and how any damages should be calculated. The case is therefore an important test of emerging AI copyright rules, not a final determination that Anthropic is liable.

Source: The Verge AI