Sony Music Publishing and Warner Music Group's publishing arm filed a federal lawsuit against Anthropic on Friday, accusing the AI company of pirating hundreds of copyrighted song compositions to train its Claude models. The labels are seeking damages of up to $150,000 per infringed work and an injunction to block further use - a claim that could add up to tens of millions of dollars given the scale of the alleged infringement.
The complaint, lodged in California federal court, names lyrics and sheet music from The Beatles, Taylor Swift, Michael Jackson, and hundreds of other artists. Sony and Warner argue Anthropic obtained the material through torrent downloads and that Claude can reproduce copyrighted lyrics "verbatim" when prompted.
The mounting legal pressure on AI training data
This lawsuit extends a pattern of copyright enforcement against generative AI companies. Universal Music Group sued Anthropic in 2023 and again earlier this year over similar claims - both cases remain active. The broader wave includes actions from authors, publishers, news outlets, and music labels challenging the ingestion of protected works without licensing agreements.
Anthropic became the first AI company to settle one of these disputes last year, paying $1.5 billion to resolve a class action brought by a group of authors. That settlement has not halted further litigation. Sony and Warner referenced it directly in their complaint: "Anthropic clearly considers that to be just the cost of doing business given that its entire business model continues to be built on copyright theft."
Anthropic's defense strategy
Anthropic's response signals it will lean on the fair use doctrine. A company spokesperson described the lawsuit as "the third lawsuit from the same lawyers, recycling allegations from cases already before the courts." The spokesperson said Anthropic will defend itself "robustly" and pointed to the judge's reasoning in the authors' class action as support for its position that AI training constitutes fair use of copyrighted material.
The labels pushed back on the deterrence value of the earlier settlement, noting that "$1.5 billion is obviously not a large enough settlement to deter infringing conduct by a company that has parlayed such mass infringement into a staggering 2-trillion-dollar valuation."
Market substitution claims and training methods
Beyond the unauthorized copying, the complaint alleges Claude was trained to "generate vast quantities of purportedly 'new' AI-generated song lyrics, which compete with Music Publishers' legitimate copyrighted works as harmful market substitutes." This argument targets not just the input side of training but the output - positioning AI-generated lyrics as a direct commercial threat to the publishers' catalogs.
The torrent allegation adds a layer of willfulness to the claims. If proven, it could undercut arguments that Anthropic's data collection was incidental or conducted in good faith through publicly available web scraping.
Why this matters for legal professionals
This case will test the boundaries of fair use in AI training for copyrighted creative works - an area where precedent remains thin. For in-house counsel and litigation teams tracking AI for Legal developments, the outcome could shape licensing norms, settlement benchmarks, and risk assessment for companies building or deploying generative AI tools. The $150,000 statutory damages per work, combined with the torrent allegations, also raise the stakes for discovery and potential exposure. Paralegals and legal researchers involved in document review for AI-related cases may find the evolving case law relevant to their work - resources like the AI Learning Path for Paralegals cover the intersection of AI systems and legal workflows that these disputes increasingly affect.
Your membership also unlocks:
