Anthropic’s $1.5 billion copyright settlement has received final court approval as a remedy for claims tied to how the company acquired and retained pirated books for alleged training of its Claude artificial intelligence models.
US District Judge Araceli Martínez-Olguín approved the non-reversionary settlement on Monday, entering final judgment and dismissing the class action with prejudice. The court found the agreement fair, reasonable, and adequate after reviewing submissions from 54 class members and potential class members.
The fund covers 482,460 books contained in versions of the Library Genesis and Pirate Library Mirror databases downloaded by Anthropic. Eligible class members include legal or beneficial owners of qualifying reproduction rights for works that met specified US copyright registration requirements.
Sponsored · Cambria Gold Mines
As of April 16, claims had been submitted for 440,490 works, representing 91.3% of the settlement list. The court estimated payments at approximately $3,000 per work before applicable deductions and divisions among rightsholders.
Valid claimants will receive pro rata shares of the net fund. Payments may be divided between authors, publishers, and other copyright owners according to publishing agreements or default allocation rules. Remaining money is expected to be redistributed to participating class members when economically practical.
Authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson brought the lawsuit in 2024, alleging that Anthropic had downloaded millions of unauthorized books and used the material in developing its AI systems.
In June 2025, then-US District Judge William Alsup ruled that Anthropic’s use of books to train Claude and its predecessor models was “exceedingly transformative” and qualified as fair use. The court also found that Anthropic could lawfully digitize print books it had purchased when the digital copy replaced the physical copy in its internal library.
Sponsored · Silver47 Exploration Inc.
Anthropic did not receive the same protection for books downloaded from pirate repositories. Alsup ruled that creating a permanent, general-purpose digital library from pirated copies was not fair use, even when some of those books were later selected for model training.
The distinction means the $1.5 billion payment resolves liability associated with unauthorized acquisition and retention, while leaving the court’s favorable ruling on AI training intact.
“We reached this settlement in 2025, after the court’s landmark ruling that training AI on books is fair use under copyright law,” Anthropic deputy general counsel Aparna Sridhar said, cited by a Reuters report.
Anthropic is privately held and backed by Amazon and Alphabet. The company denied wrongdoing when it agreed to the settlement.
In the recent settlement, Martínez-Olguín awarded class counsel $101.56 million, equal to nearly 6.8% of the settlement fund. The lawyers had most recently requested $187.5 million after initially seeking $300 million.
The court also approved $2.64 million in reimbursed expenses and created an $18.22 million reserve for anticipated administration and litigation costs. Any payment from that reserve will require further court approval.
Each of the three class representatives received a $15,000 service award. The plaintiffs had requested $50,000 each, but the court found that amount unreasonable without evidence that the representatives faced retaliation or comparable risks.
Ten percent of the attorney fee award will remain withheld until the parties submit a post-distribution accounting. The judge retained authority to reduce that portion if final documented legal work falls below counsel’s projections.
Only 350 valid opt-outs covering 1,802 works were submitted by the deadline. Some authors and publishers that excluded themselves have filed separate lawsuits against Anthropic, according to Reuters. Those cases remain pending.
