The Anthropic copyright settlement is a $1.5 billion class-action agreement resolving claims that the company downloaded millions of pirated books from shadow libraries to train its Claude AI models. Authors, publishers, and other rights holders whose books appear on the settlement’s Works List stand to receive roughly $3,000 to $3,100 per eligible work. A federal judge granted preliminary approval in September 2025 and held a final fairness hearing on May 14, 2026; final approval and initial payments are expected in late 2026.1Publishing Perspectives. Anthropic Settlement Appears to Cruise Through Its Final Fairness Hearing2Authors Guild. Anthropic Settlement Update: 91 Percent of Books Claimed
What Anthropic Was Accused Of
Authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson filed Bartz v. Anthropic PBC in August 2024 in the U.S. District Court for the Northern District of California.3Courthouse News Service. Authors, Publishers Near Final Approval of $1.5 Billion Anthropic Copyright Settlement The complaint did not attack AI training as such. It attacked how Anthropic got the books it trained on.
Between 2021 and 2022, Anthropic pulled its training library from two pirate repositories known as shadow libraries. Cofounder Ben Mann downloaded the Books3 dataset in February 2021, roughly five million books from Library Genesis (LibGen) in June 2021, and the company downloaded another five million from the Pirate Library Mirror (PiLiMi) in July 2022.4Publishers Weekly. Federal Judge Rules AI Training Is Fair Use in Anthropic Copyright Case The court found the company knew the sources were pirated.5Copyright Alliance. Bartz v. Anthropic Order
Internal documents unsealed in January 2026 showed Anthropic executives referred to a parallel physical book-scanning operation as “Project Panama,” which a planning document described as “our effort to destructively scan all the books in the world.”6Washington Post. Anthropic AI Scan Destroy Books Judge William Alsup wrote in a summary judgment opinion that although Anthropic had “many places” to buy books, it “preferred to steal them” to avoid what CEO Dario Amodei internally called a “legal/practice/business slog.”4Publishers Weekly. Federal Judge Rules AI Training Is Fair Use in Anthropic Copyright Case
Why Anthropic Settled Instead of Going to Trial
In June 2025, Judge Alsup split the case in half on summary judgment. He ruled that training AI models on copyrighted books is “exceedingly transformative” and qualifies as fair use under Section 107 of the Copyright Act. He also found that buying print books, disassembling them, and scanning them into a digital research library was fair use.5Copyright Alliance. Bartz v. Anthropic Order7Fortune. AI Training Is Fair Use Federal Judge Rules Anthropic Copyright Case
The piracy was a different matter. Applying the four fair use factors to Anthropic’s downloads from LibGen and PiLiMi, Alsup found “every factor points against fair use.” The pirated copies were a “permanent, general-purpose resource,” not a transformative use. He doubted “any accused infringer could ever meet its burden of explaining why downloading source copies from pirate sites that it could have purchased or otherwise accessed lawfully was itself reasonably necessary to any subsequent fair use.”7Fortune. AI Training Is Fair Use Federal Judge Rules Anthropic Copyright Case
Alsup ordered the piracy claims to trial on damages. With statutory damages of up to $150,000 per willfully infringed work and more than seven million downloaded copies at issue, observers described Anthropic’s exposure as “bet-the-company” liability potentially exceeding $70 billion.8NPR. Anthropic Settlement Authors Copyright AI The parties announced the $1.5 billion settlement in late August 2025.9Bloomberg Law. Record $1.5 Billion AI Copyright Pact Sets Bar for OpenAI, Meta
Which Books Qualify
The certified class covers the legal or beneficial copyright owners of books that meet three tests: the book was downloaded by Anthropic from LibGen or PiLiMi, it has an ISBN or Amazon identification number, and it was registered with the U.S. Copyright Office within five years of publication.10Anthropic Copyright Settlement. FAQ Authors, publishers, literary estates, and exclusive licensees are all eligible. The resulting Works List contains 482,460 eligible titles out of the roughly seven million copies Anthropic downloaded.11Society of Authors. Anthropic List of Stolen Works Published
Books that were not registered with the Copyright Office within five years of publication, foreign works without U.S. registration, and titles Anthropic obtained only through print-scanning are outside the class. If a book was pirated but is not on the Works List, it does not qualify for payment under this settlement.
How Much Each Author or Publisher Gets
The estimated payout is about $3,000 to $3,100 per eligible work, distributed equally across every work with a valid claim. The settlement fund is non-reversionary, so unclaimed money does not go back to Anthropic.3Courthouse News Service. Authors, Publishers Near Final Approval of $1.5 Billion Anthropic Copyright Settlement
When both an author and a publisher hold rights in the same book, the default split is 50/50 for trade and university press titles. Self-published authors, and authors whose rights have reverted, receive the full per-work amount. Educational texts have no default split and require documentation of contract terms.12Authors Guild. What Authors Need to Know About the Anthropic Settlement
Anthropic is paying in four installments: $300 million was deposited in October 2025, another $300 million is due in April 2026, $450 million by September 2026, and a final $450 million by September 2027. The third and fourth installments accrue interest from September 25, 2025.2Authors Guild. Anthropic Settlement Update: 91 Percent of Books Claimed
What Anthropic Must Do Beyond Paying
Within 30 days of final judgment, Anthropic must destroy all copies of works originating from LibGen and PiLiMi, including derivative copies, and certify the destruction in writing to class counsel.3Courthouse News Service. Authors, Publishers Near Final Approval of $1.5 Billion Anthropic Copyright Settlement
The settlement is narrower than it sounds. It does not grant Anthropic any license to use copyrighted works going forward, and it only releases claims for conduct before August 26, 2025. Claims about infringing AI outputs — for example, that Claude reproduces protected text — remain actionable.13Copyright Alliance. Participating in the Bartz v. Anthropic Settlement
Where the Case Stands Now
By the May 2026 fairness hearing, 447,576 of the 482,460 eligible works had been claimed, a claim rate of 92.77%.1Publishing Perspectives. Anthropic Settlement Appears to Cruise Through Its Final Fairness Hearing14Ars Technica. Authors Fight for Higher Payouts From Anthropic’s $1.5B Copyright Settlement15Authors Alliance. Bartz v. Anthropic Settlement Update
Judge Araceli Martínez-Olguín, who took over the case after Judge Alsup went inactive, held a 75-minute hearing on May 14, 2026, in San Francisco.16Authors Alliance. Bartz v. Anthropic Updated Opt-Out and Objection Dates and a New Judge Seven objectors testified. The judge did not rule from the bench. She set a May 21, 2026, deadline for class counsel to respond and for Anthropic to state its position on late opt-outs.17Publishers Weekly. Little Drama at Anthropic’s Settlement Hearing The Authors Guild has advised claimants that payments will likely begin in late fall 2026 at the earliest, depending on any appeals.2Authors Guild. Anthropic Settlement Update: 91 Percent of Books Claimed
The Authors Who Walked Away
A group of authors chose to opt out and sue on their own. Journalist John Carreyrou, author of Bad Blood, led six writers who filed Carreyrou v. Anthropic PBC et al. in the Northern District of California in December 2025 against Anthropic, OpenAI, Google, Meta, xAI, and Perplexity AI. The suit is not a class action. The plaintiffs argue that $3,000 per work is just 2% of the statutory maximum, and they are seeking $150,000 per work against each of the six defendants.18Bloomberg Law. OpenAI, Anthropic, xAI Hit With Copyright Lawsuit From Writers19Publishers Weekly. Authors File New Lawsuit Against AI Companies Seeking More Money A separate group of 25 class members who opted out filed their own suit against Anthropic on May 13, 2026.14Ars Technica. Authors Fight for Higher Payouts From Anthropic’s $1.5B Copyright Settlement
What It Signals for Other AI Cases
The settlement itself sets no binding legal precedent. Anthropic denied wrongdoing, and the infringement question was never tried.20Authors Alliance. Bartz v. Anthropic Settlement FAQ Judge Alsup’s underlying ruling does carry weight, at least in the Northern District of California: training AI on lawfully obtained copies can be fair use; building the training set from pirate sites is not.
More than 50 AI copyright lawsuits are pending in the United States against companies including OpenAI and Meta.13Copyright Alliance. Participating in the Bartz v. Anthropic Settlement The $3,000 per-work figure has been described as a potential floor that could shape future negotiations.9Bloomberg Law. Record $1.5 Billion AI Copyright Pact Sets Bar for OpenAI, Meta The Authors Guild said the settlement “sends a clear signal to AI companies that infringement of authors’ rights comes at a steep price,” while noting it “respectfully disagrees” with the fair use finding and continues to press that argument in its own class action against OpenAI in the Southern District of New York.21Authors Guild. Authors Guild Statement on Approval of Anthropic Settlement