Music Publishers Escalate Their Copyright Fight Against Anthropic

Sony Music Publishing, Warner Chappell and other publishers accuse Anthropic of using pirated books, lyrics and sheet music to build Claude. Anthropic says the allegations recycle prior claims and maintains AI training is transformative fair use.
Music Publishers Escalate Their Copyright Fight Against Anthropic

Music Publishers Escalate Their Copyright Fight Against Anthropic
Music publishers portray Anthropic’s AI boom as built on uncompensated creative work, while the company casts the new case as a recycled legal attack on a technology it says is protected by fair use.

Sony Music Publishing, Warner Chappell and other publishers filed suit in federal court in Northern California against Anthropic and its co-founders, Dario Amodei and Benjamin Mann. The complaint alleges that Claude was developed using thousands of copyrighted compositions obtained through torrenting, scraping and unauthorized downloads.

The publishers’ case reaches back to July 2021, when it says Mann personally used BitTorrent to acquire millions of pirated books from Library Genesis. It further alleges that Anthropic staff later sought material from the Pirate Library Mirror, including books containing sheet music and lyrics. The filing characterizes the conduct as a “brazen campaign of illegally torrenting, scraping, and downloading copyrighted works on a massive scale.”

The suit also claims Anthropic copied lyrics from sources including MusixMatch and LyricFind, and names songs such as “Ain’t No Mountain High Enough,” “Livin’ On a Prayer,” “September,” “Hallelujah” and Taylor Swift’s “Paper Rings” as examples found in the company’s training data. Publishers seek damages for tens of thousands of works — potentially up to $150,000 per work — as well as penalties for allegedly stripped copyright-management information.

Their central argument is not simply that Claude learned from protected material, but that its outputs can reproduce lyrics, blend them into newly generated songs and compete with writers whose work allegedly helped train it. They want an injunction and greater disclosure of Anthropic’s training data, methods and model capabilities.

The lawsuit follows Anthropic’s $1.5 billion settlement with authors over pirated books. That earlier litigation drew a distinction between using copyrighted works in AI development and obtaining them through piracy; this music case presses the market-harm question more directly.

Anthropic rejects the publishers’ framing. “This is the third lawsuit from the same lawyers, recycling allegations from cases already before the courts,” a spokesperson said, adding that generative-AI training is “a transformative fair use” and that the company will defend itself.

Continue reading https://foxvector.com/stories/01a059a9-407b-0d49-7136-1943cc6a3d70

Write a comment