ArtistDirect News
Sony Music Publishing and Warner Chappell File $150,000-Per-Song Lawsuit Against Anthropic Over Claude Training Data
Sony Music Publishing and Warner Chappell Music filed a federal complaint on August 29, 2026, accusing Anthropic, the maker of the Claude chatbot, of illegally using tens of thousands of copyrighted songs to train its large‑language models. The 48‑page filing was lodged in the U.S. District Court for the Northern District of California and names Anthropic, CEO Dario Amodei and co‑founder Benjamin Mann as defendants.
The plaintiffs allege that Anthropic acquired lyrics, sheet music and related text through unauthorized channels, including torrent sites, the Common Crawl archive and the pirate database Library Genesis. The lawsuit lists specific tracks that were allegedly used without permission, such as Mariah Carey’s “All I Want for Christmas Is You,” Survivor’s “Eye of the Tiger,” Marvin Gaye’s “Ain’t No Mountain High Enough,” and songs by Taylor Swift, The Beatles and Bon Jovi.
According to the complaint, Anthropic not only used the copyrighted material to train its models but also allowed Claude to reproduce verbatim lyrics in response to user prompts. The plaintiffs argue that this practice directly competes with the original works and that Anthropic removed identifying metadata from the songs during processing, preventing proper credit and attribution to the original songwriters.
Sony and Warner are seeking a jury trial and are demanding statutory damages of up to $150,000 for each infringed composition and $25,000 for each instance of altered metadata. Because the complaint references tens of thousands of tracks, the potential liability could reach billions of dollars.
Anthropic’s legal history includes a $1.5 billion settlement with authors over a 2025 class‑action lawsuit that alleged the company had downloaded pirated books for training. In its response to the current suit, an Anthropic spokesperson said the company believes training generative AI models is a transformative use that falls under fair‑use provisions of copyright law. The spokesperson also stated that Anthropic intends to defend itself vigorously in court.
The lawsuit illustrates a growing tension between major record labels and AI developers over the sourcing of training data. While some AI firms have entered into licensing agreements with publishers, others have relied on publicly available or unlicensed datasets. Sony and Warner’s filing signals that the industry may pursue more aggressive legal action if it perceives that AI companies are not obtaining proper permissions.
The case also highlights the broader debate about the limits of fair use in AI training. Courts have yet to issue a definitive ruling on whether large‑language models that reproduce copyrighted text qualify as transformative. The outcome of this lawsuit could set a precedent for how music and other creative works are treated in AI training pipelines.
At present, the lawsuit remains pending. Both parties have not yet indicated whether they will negotiate a settlement or proceed to trial. The case will be monitored closely by industry observers, legal analysts, and AI developers as it may shape the future of data sourcing and copyright compliance in the rapidly evolving AI landscape.
The plaintiffs allege that Anthropic acquired lyrics, sheet music and related text through unauthorized channels, including torrent sites, the Common Crawl archive and the pirate database Library Genesis. The lawsuit lists specific tracks that were allegedly used without permission, such as Mariah Carey’s “All I Want for Christmas Is You,” Survivor’s “Eye of the Tiger,” Marvin Gaye’s “Ain’t No Mountain High Enough,” and songs by Taylor Swift, The Beatles and Bon Jovi.
According to the complaint, Anthropic not only used the copyrighted material to train its models but also allowed Claude to reproduce verbatim lyrics in response to user prompts. The plaintiffs argue that this practice directly competes with the original works and that Anthropic removed identifying metadata from the songs during processing, preventing proper credit and attribution to the original songwriters.
Sony and Warner are seeking a jury trial and are demanding statutory damages of up to $150,000 for each infringed composition and $25,000 for each instance of altered metadata. Because the complaint references tens of thousands of tracks, the potential liability could reach billions of dollars.
Anthropic’s legal history includes a $1.5 billion settlement with authors over a 2025 class‑action lawsuit that alleged the company had downloaded pirated books for training. In its response to the current suit, an Anthropic spokesperson said the company believes training generative AI models is a transformative use that falls under fair‑use provisions of copyright law. The spokesperson also stated that Anthropic intends to defend itself vigorously in court.
The lawsuit illustrates a growing tension between major record labels and AI developers over the sourcing of training data. While some AI firms have entered into licensing agreements with publishers, others have relied on publicly available or unlicensed datasets. Sony and Warner’s filing signals that the industry may pursue more aggressive legal action if it perceives that AI companies are not obtaining proper permissions.
The case also highlights the broader debate about the limits of fair use in AI training. Courts have yet to issue a definitive ruling on whether large‑language models that reproduce copyrighted text qualify as transformative. The outcome of this lawsuit could set a precedent for how music and other creative works are treated in AI training pipelines.
At present, the lawsuit remains pending. Both parties have not yet indicated whether they will negotiate a settlement or proceed to trial. The case will be monitored closely by industry observers, legal analysts, and AI developers as it may shape the future of data sourcing and copyright compliance in the rapidly evolving AI landscape.