Sony Music, Warner Chappell Sue Anthropic Over Training Data

Sony Music Publishing and Warner Chappell sued Anthropic on Friday, seeking up to $150,000 for each of "tens of thousands" of copyrighted works. The publishers also named co-founders Dario Amodei and Benjamin Mann as individual defendants, alongside Anthropic itself.

The suit was filed in the U.S. District Court for the Northern District of California and was first reported by Music Business Worldwide. If a court found for the publishers and awarded the maximum, The Verge reports the damages could reach several billion dollars.

The damages math

Two figures drive the total. The publishers are asking for up to $150,000 per work across tens of thousands of works, plus up to $25,000 for each instance in which identifiable copyright data was stripped from a work.

That second number is the one worth reading twice. It is charged per instance rather than per work, which means the count scales with how many times identifying metadata was removed during data preparation — not with the size of the catalog. For a training corpus assembled at scale, those two multipliers compound in different directions.

The complaint names specific songs it says were found in Anthropic's training data, including Marvin Gaye and Tammi Terrell's "Ain't No Mountain High Enough," Bon Jovi's "Livin' On a Prayer," Earth, Wind & Fire's "September," Leonard Cohen's "Hallelujah," and Taylor Swift's "Paper Rings."

What the publishers allege

The filing opens by describing the case as an action "to hold accountable the culprits behind one of the largest and most blatant ongoing thefts of intellectual property in history," and accuses Anthropic and its founders of "a brazen campaign of illegally torrenting, scraping, and downloading copyrighted works on a massive scale."

Three specific allegations sit underneath that language, per The Verge's reading of the complaint:

• Mann used BitTorrent to download over five million pirated books.
• Employees downloaded at least two million more from Pirate Library Mirror.
• Anthropic scraped lyrics from sites including MusixMatch and LyricFind — services that themselves paid to license content from the labels.

These are allegations in a filed complaint, not findings. Anthropic has not answered them in court.

The third item is the structurally novel one. Books and torrents were already the terrain of the earlier cases; scraping licensed intermediaries adds a second acquisition path, one where the content sat behind a commercial license the plaintiffs say they were paid for.

Why the Bartz ruling shapes this case

Anthropic was ordered to pay $1.5 billion in the landmark Bartz v. Anthropic case, TechCrunch reports, after a judge ruled that while it was legal for the lab to use copyrighted works, it was not legal to acquire that content through piracy.

That distinction is the hinge. Under it, the contested question is no longer whether training on copyrighted material is permitted — it is how the material got onto the disk. A complaint built almost entirely on acquisition conduct is aiming at the part of the earlier ruling that went against Anthropic, and stepping around the part that went in its favor.

TechCrunch notes the two cases make similar arguments but differ in scope: this one is broader and builds off the others, accusing Anthropic of "flagrant piracy" through illegal torrenting to obtain millions of copies of books, including those containing lyrics and sheet music.

The continuity is not only legal. Some of the lawyers behind this suit also represent Concord Music Group and Universal Music Group in a case filed in January, and led the Bartz case. This is a team relitigating a theory that already produced the $1.5 billion Bartz outcome.

Anthropic's response

"We disagree with the publishers' claims and we intend to defend ourselves robustly in court," an Anthropic spokesperson wrote in a statement to TechCrunch.

The suit lands on a company that is already carrying a crowded docket. Beyond Bartz, The Verge notes Anthropic has faced multiple lawsuits from Universal Music Group, Concord, and ABKCO, plus separate suits from BMG and Round Hill Music.

One clarification worth carrying

The plaintiff is Sony Music Publishing, not Sony Music Entertainment — The Verge appended that clarification on August 30. It matters for reading the complaint, which is built around lyrics scraped from licensing services and books containing lyrics and sheet music, rather than around sound recordings.

What to watch

The first real signal will be Anthropic's answer to the complaint, and specifically whether it contests the acquisition allegations on the facts or attacks them on legal grounds. The second is the per-instance metadata claim: if a court lets that theory run at scale, the exposure calculation for every lab that stripped identifying data during preprocessing changes shape, regardless of how the training-use question is eventually settled.

For teams building on these models, nothing changes operationally today. What is worth tracking is the trend line: two of the three questions in play — where training data came from, and what was done to its metadata — are questions about pipelines, not about model outputs. That is a compliance surface engineering teams can actually be asked about.