Sony and Warner Music File Billion Dollar Copyright Lawsuit Against Anthropic

Some of the largest music publishers on the planet have initiated sweeping legal action against the artificial intelligence startup Anthropic. Companies tied to Sony Music Publishing and Warner Chappell Music submitted a formal complaint in a Northern California federal court on Friday. The legal filing accuses the tech firm and its executives of orchestrating massive, unauthorized data harvesting operations to build their popular Claude algorithms.

The plaintiffs are demanding an initial jury trial and are asking for statutory damages up to $150,000 for each musical work they claim the startup unlawfully copied. They also seek an extra penalty of $25,000 for every instance where copyright management details were removed from the text. The legal teams representing the labels described the startup's actions as an ongoing and blatant theft of protected intellectual property.

Allegations of Widespread Torrenting

The lawsuit details a multi-pronged approach that Anthropic allegedly used to gather training data. According to the court documents, CEO Dario Amodei and co-founder Benjamin Mann personally directed efforts to acquire millions of pirated books and texts using the BitTorrent protocol. The complaint claims that Mann downloaded at least five million pirated files from Library Genesis in June 2021. Shortly after, other employees allegedly secured another two million files from the Pirate Library Mirror.

The labels maintain that these massive data dumps contained hundreds of songbooks and sheet music collections holding their copyrighted compositions. Because torrenting works by simultaneously downloading and sharing pieces of files with other users, the publishers argue that the startup also violated their exclusive distribution rights.

Beyond the peer-to-peer sharing networks, the music companies accuse the startup of scraping words from officially licensed websites like MusixMatch and LyricFind without permission. The complaint also mentions a destructive scanning operation where millions of physical books were digitized and then destroyed.

Generating Verbatim Copies

The tech industry closely watches intellectual property disputes involving automated software. In this case, the publishers claim that Claude frequently memorizes the lyrics fed into its system during the training phase. When prompted by a user, the algorithm can easily reproduce nearly identical versions of popular songs like "Ain't No Mountain High Enough" and "All I Want for Christmas is You".

While the tech firm added safety filters following earlier legal disputes with other music groups, the new filing argues that people can bypass those restrictions by simply asking the software again. The record labels believe that the ability to instantly generate these texts creates a direct market competitor. If consumers can find lyrics through a chatbot instead of a licensed service, the publishers argue it directly hurts the royalty pools used to pay songwriters.

The financial stakes attached to this dispute could easily reach the multi-billion dollar range. This confrontation represents a growing tension between creative industries and technology companies racing to build more capable algorithms. Both sides are now preparing for a lengthy legal fight that could set strict boundaries on how software developers collect training data in the future.