🔍 Read the full analysis: What Is Alleged—and Proven—About Anthropic, Claude, And Books? on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
A federal judge ruled in June 2025 that Anthropic’s use of books to train Claude was fair use, but that downloading millions of pirated copies to build a library was not. Anthropic later agreed to a $1.5 billion settlement over the downloads; new lawsuits by authors who opted out have not been decided. The settlement and court record do not establish that pirated books were used to train commercially released Claude models.
A federal judge ruled in June 2025 that Anthropic’s training of Claude on books was fair use, while finding that downloading more than seven million pirated book copies to create a permanent library was not. Anthropic later agreed to a $1.5 billion settlement covering claims tied to those downloads; separate lawsuits from authors who opted out remain unresolved.
The decision came in Bartz v. Anthropic, a copyright case filed by three authors in August 2024. In a June 23, 2025 summary-judgment order, U.S. District Judge William Alsup distinguished between training Claude and obtaining books. He found that training on books was transformative fair use, and also found that Anthropic’s purchase and scanning of millions of used print books was fair use. But he ruled against retaining pirated digital copies from collections including Books3, Library Genesis and the Pirate Library Mirror.
The order described 196,640 books from Books3, at least five million copies from LibGen and at least two million from PiLiMi. Alsup wrote that the record involved “over seven million copies of books.” The court’s finding on piracy was made at the summary-judgment stage, before a jury trial on remaining questions such as damages. The case settled before that trial, so there was no jury verdict determining damages.
The settlement, finally approved in July 2026, provides $1.5 billion for owners of roughly 500,000 registered books identified as downloaded from LibGen or PiLiMi. That works out to about $3,000 per book before fees. Anthropic agreed to destroy the downloaded collections and copies, subject to legal-hold obligations. The agreement resolves past claims through August 25, 2025; it does not release claims about outputs generated by Claude.
From download to IPO
Five years, one ruling, no trial
The piracy happened in 2021 and 2022. The only ruling came in 2025. The new lawsuits are years away from a verdict.
- Jun 2021 · Jul 2022DownloadsAt least 5 million copies from LibGen (2021), at least 2 million from PiLiMi (2022); 196,640 from Books3 earlier.
- Aug 2024Bartz v. Anthropic filedThree authors sue Anthropic, on behalf of a class of book owners.
- 23 Jun 2025Judge Alsup's orderTraining on books: fair use. Scanning bought books: fair use. Pirated library copies: not fair use, to trial.
- Sep 2025$1.5 billion settlementNo trial. Anthropic denies wrongdoing; release covers conduct up to 25 Aug 2025 only.
- 13 + 15 May 2026Opt-out suitsCruz (28 plaintiffs) and Kwon (incl. Chang-rae Lee) join Cambronne and two publisher suits.
- Jul 2026Final approvalSettlement approved. Anthropic: no opt-out deal "above the Bartz claim value".
- 24 Sep 2026Motion to dismissAnthropic: claims too late, no infringing output shown, many works already released.
- 26–27 Oct 2026NowS-1 and the headlineIPO filing cites Bartz as a risk; The Stanford Daily names Stanford authors.
- 17 Dec 2026ScheduledHearingCourt hears the motion to dismiss.
- May 2027 · Jan 2028ScheduledDiscovery, then trial settingFact discovery closes 13 May 2027; trial-setting conference 11 Jan 2028.
Sources: Bartz v. Anthropic, order of 23 June 2025; settlement FAQ; joint case statement, 27 Aug 2026; Anthropic letter, 27 Jul 2026; case management order, 14 Sep 2026; Anthropic S-1, 26 Oct 2026; The Stanford Daily, 27 Oct 2026. Scheduled dates can move.
Powered by Thorsten Meyer AI
Copyright Liability and Claude Training
The distinction matters because the case is often summarized as a court finding that Anthropic “illegally trained” Claude on books. That is not what Judge Alsup decided. His order treated model training and pirated acquisition as separate acts: training was fair use under the facts before him, while downloading and keeping pirated copies was not.
The settlement gives eligible rights holders a substantial payment and requires deletion of the specified pirated libraries, but it is not a general licence for AI companies to use copyrighted books. Nor does it settle every dispute over the use of books in AI systems. Authors who opted out have brought separate cases, and their claims are still before the courts.
For readers, writers and publishers, the record illustrates how copyright questions around AI can turn on the source and handling of training material, not only on what a model generates. For companies developing AI, the financial outcome also shows that obtaining large collections without permission can create significant legal exposure even when a court finds the later training process fair use.
Products Worth Considering
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
Jackery Solar Generator 1000 v2 with 200W Solar Panel
Bose S1 Pro+ Portable PA Speaker
As an affiliate, we earn on qualifying purchases.
From Bartz Ruling to New Cases
After the 2025 order, Anthropic and the authors reached the settlement covering the pirated-book claims. The settlement releases only specified past conduct, not claims about Claude’s outputs, and requires destruction of the LibGen and PiLiMi copies covered by the agreement. Anthropic denies wrongdoing and represented to the settlement court that neither those datasets nor portions of them were in the training corpus of any commercially released large language model.
That representation should not be conflated with the judge’s findings about the earlier acquisition and use of material. The source account says the 2025 record described portions of the pirated collections being used to train some Anthropic models, while the settlement representation addresses the training corpus of commercially released models. The court did not hold that pirated books were used to train commercially released Claude models.
In 2026, authors who did not join the settlement filed new lawsuits. The Stanford Daily reported that books by several Stanford-affiliated authors were among the works affected, and quoted novelist Tobias Wolff criticizing unpaid use of authors’ work. Those reports are not a court ruling. The source account says Chang-rae Lee is listed as a plaintiff in one of the new actions, while it could not verify Wolff’s name among the plaintiffs in five related cases as of August 2026. It also says the reported book list was not independently checked against the settlement database.
““over seven million copies of books””
— Judge William Alsup, describing the pirated collections in the 2025 order
Questions the Court Has Not Resolved
The new cases brought by authors who opted out have not been decided. A hearing on Anthropic’s motion to dismiss is scheduled for December 17, 2026, according to the source account. The claims, any defenses and the court’s eventual rulings remain unresolved.
The available material also does not establish whether every book named in media coverage appears in the settlement database, or whether Wolff is a plaintiff in a later amendment or another action. The settlement’s scope is limited to the released past conduct; it does not resolve claims over Claude’s outputs. Anthropic’s statements about commercial-model training remain its representations to the settlement court, rather than a separate judicial ruling on every model and dataset.
The next stated milestone is the December 17, 2026 hearing on Anthropic’s motion to dismiss the opt-out authors’ claims. The court may then decide whether some or all of those claims can proceed; the hearing itself will not necessarily resolve the underlying copyright questions.
The settlement also schedules payments, including $450 million by September 27, 2027, under the terms described in the source account. Further court filings may clarify which authors and books are included in the remaining cases. Until those proceedings advance, the established record remains bounded: the 2025 judge found training fair use and pirated downloading unlawful, while later claims about other conduct have yet to be adjudicated.
Key Questions
Did a judge rule that Anthropic illegally trained Claude on books?
No. In June 2025, Judge William Alsup found training Claude on books was fair use under the facts before him. He ruled that downloading and keeping pirated copies to build a library was not fair use.
What does Anthropic’s $1.5 billion settlement cover?
It covers claims tied to pirated books downloaded from LibGen and PiLiMi for roughly 500,000 registered books. The agreement requires destruction of the specified copies and releases certain past claims, but it does not release claims about Claude’s outputs.
Did the court find pirated books were used to train commercial Claude models?
The settlement record includes Anthropic’s representation that neither the LibGen and PiLiMi datasets nor portions of them were in the training corpus of any commercially released large language model. That is Anthropic’s representation; the court’s ruling did not establish that pirated books were used to train commercially released Claude models.
Are the lawsuits by authors who opted out resolved?
No. The new cases remain pending, and a hearing on Anthropic’s motion to dismiss is scheduled for December 17, 2026, according to the source account.
Source: ThorstenMeyerAI.com
