Anthropic AI, producer of the Claude generative AI platform, has agreed to settle a copyright infringement lawsuit with authors who alleged – and confirmed – that Anthropic AI had used their content to train Claude without licensing it.
However, Anthropic also ingested books that it had obtained legally into their platform. The judge in the case said that “The training use was a fair use….The use of the books at issue to train Claude and its precursors was exceedingly transformative.” However, the judge drew the line at the use of pirated content.
“To the best of our knowledge, this is the largest known copyright settlement in American history. A certified class of legal and beneficial owners of hundreds of thousands of copyrighted works will recover at least $1,500,000,000 dollars through a non-reversionary settlement fund,” according to a court document dated September 5, 2025.
If the settlement in case #3:24-cv-05417 is approved by the US District Court for the Northern District of California, “Anthropic will pay the Class at least $1.5 billion dollars. With around 500,000 works in the Class, this amounts to an estimated gross recovery of $3,000 per Class Work…” (plus interest accrued while funds are in escrow before they are distributed).
Removing training content
Anthropic PBC, will delete the books it downloaded from Library Genesis (“LibGen”) and Pirate Library Mirror (“PiLiMi”), subject to certain legal preservation obligations and court orders. Both are recognized sources of unlicensed content used for training by multiple generative AI platforms.
Further background
“Plaintiffs’ core allegation is that Anthropic committed largescale copyright infringement by downloading and commercially exploiting books that it obtained from allegedly pirated datasets… Plaintiffs secured critical victories for the Class. These included: (i) uncovering Anthropic’s use of material from Library Genesis (“LibGen”) and Pirate Library Mirror (“PiLiMi”); (ii) the partial denial of Anthropic’s summary judgment motion, which found Anthropic’s use of pirated material “inherently, irredeemably infringing,”
Due diligence
“Over the course of the case, Plaintiffs took and defended 20 depositions, reviewed hundreds of thousands pages of documents, conducted inspections of at least 3 TB of training data, and litigated discovery motions. And after the Court’s Class Certification Order, Class Counsel—in conjunction with their experts—worked day and night to assemble the Works List by matching millions of records of Anthropic’s downloads to U.S. Copyright Office registration records.”
Further reading:
Consent motion for settlement. Document #362. Bartz v Anthropic PBC, Case 3:24-cv-05417. US District Court for the Northern District of California. via Courtlistener.com
Class Action Complaint. Document #1. Bartz v Anthropic PBC, Case 3:24-cv-05417. US District Court for the Northern District of California. via Courtlistener.com
Court documents. Bartz v Anthropic PBC, Case 3:24-cv-05417. US District Court for the Northern District of California. via Courtlistener.com
Copyright Alliance CEO issues statement on ‘Bartz v. Anthropic’ case settlement. Press release. September 8, 2025. Copyright Alliance.
Anthropic settles with authors in first-of-its-kind AI copyright infringement lawsuit. Article. September 5, 2025. by Chloe Veltman. NPR
Anthropic raises $13B Series F at $183B post-money valuation. Press release. September 2, 2025. Anthropic
Why it matters
“The settlement is a thunderclap for AI companies and anyone else attempting to copy and use pirated works from illicit sources for AI training or other uses,” said Keith Kupferschmid, CEO of the Copyright Aliance. “Because many other AI companies have also been using these so-called shadow libraries and other illicit sources to access and use pirated works for training their AI systems, the outcome of this case is likely not unique and will apply in many other cases and to many other AI companies.” he said.
While the $1.5 billion settlement seems to be an enormous sum of money in the everyday world, it is almost a rounding error in the tech universe. Just days before the Anthropic settlement was reported, Anthropic announced that it had raised a $13 billion series F funding round. This places Anthropic’s total post-money (e.g. including money raised) value at $183 billion.
Other cases are ongoing.
