🔍 Read the full analysis: Did Anthropic Train Claude On Pirated Books? What Is Alleged, What Is Proven, And What It Means For The IPO on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
A federal judge ruled in June 2025 that training Claude on books was fair use, but that Anthropic’s downloading and retention of more than seven million pirated book copies was not. Anthropic later agreed to a $1.5 billion settlement covering about 500,000 books; new lawsuits by authors who opted out remain unresolved, with a dismissal hearing scheduled for December 17, 2026.
A federal judge ruled in June 2025 that Anthropic’s use of books to train Claude was fair use, while finding that the company’s downloading and permanent storage of more than seven million pirated book copies was not. Anthropic later settled the authors’ claims for $1.5 billion; separate lawsuits filed by authors who opted out have not been decided.
The ruling came in Bartz v. Anthropic, a 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 on books and acquiring copies without permission. He found that training on books and scanning books Anthropic had bought were fair use. But he said downloading and keeping pirated copies was not. The order identified 196,640 books from Books3, at least five million from Library Genesis and at least two million from Pirate Library Mirror.
The court did not hold a trial on damages. It accepted the authors’ account of the pirated downloads for purposes of the order, but a jury was expected to decide remaining issues before the parties settled. The settlement, finally approved in July 2026, provides $1.5 billion for rights holders of roughly 500,000 registered books. That works out to about $3,000 per book before deductions. The agreement also requires Anthropic to destroy the downloaded LibGen and PiLiMi books and copies, subject to legal-hold obligations.
Anthropic denies wrongdoing. In the settlement process, the company represented that neither the LibGen nor PiLiMi datasets, nor portions of them, were included in the training corpus of any commercially released large language model. The court’s 2025 order addressed the use of pirated copies in training some models, but the settlement resolved the claims without a trial on damages. The agreement releases claims over past conduct through August 25, 2025; it does not release claims about Claude’s outputs or grant a general license for future training.
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
What the Settlement Covers
The distinction between training methods and data acquisition is central to the dispute. The judge’s fair-use finding means the ruling did not declare all AI training on copyrighted books unlawful. At the same time, it treated obtaining and retaining pirated copies as a separate copyright problem, even where the company later bought books or used material to train a model.
The settlement’s scale also makes the case important to authors and AI developers. Its $1.5 billion payment addresses registered books tied to the pirated collections, but it does not settle every dispute about AI and copyright. Authors who opted out can continue pursuing claims, and the agreement does not resolve questions about material generated by Claude.
Products Worth Considering
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
Apple iPad Air 11-inch (M2), 512GB, Wi-Fi, Space Gray
As an affiliate, we earn on qualifying purchases.
How the Cases Diverged
The dispute has generated headlines that sometimes merge different events. A report by The Stanford Daily said books by Stanford-affiliated writers, including Tobias Wolff, Abraham Verghese and Chang-rae Lee, were among the pirated copies. That identification is the newspaper’s reporting; the article’s book list has not been independently checked against the settlement database in the source account. Chang-rae Lee is listed as a plaintiff in one of the new suits. The account says it could not find Wolff among plaintiffs in five related cases listed as of August 2026.
The new cases are separate from the 2025 Bartz ruling. Authors who opted out of the settlement filed lawsuits in May 2026, and no judge has yet ruled on their allegations. Anthropic has asked for the claims to be dismissed. A hearing is scheduled for December 17, 2026. The settlement itself covers past conduct only and does not decide those opt-out claims.
Anthropic’s public offering filing is another separate source of claims circulating alongside the copyright story. According to the source account’s review of the S-1, it does not say Claude is sentient or could blackmail people. It describes test behavior resembling blackmail, notes public debate over possible AI sentience and regulation, and refers to existential concerns raised by others.
“Neither the LibGen or PiLiMi datasets, nor any portions of those datasets, were in the training corpus of any of its commercially released large language models.”
— Anthropic, in a representation described in the settlement record
What the Court Did Not Decide
The ruling established that Anthropic downloaded and retained pirated copies and found that conduct was not fair use, but damages were not decided at trial. The settlement ended the original case before a jury could resolve that question. The settlement’s compensation terms do not amount to a finding that every book in the reported lists was used in a commercially released model.
The opt-out authors’ claims remain allegations, not findings. Their case may test issues not resolved by the settlement, but the outcome and the court’s view of Anthropic’s arguments are unknown. It is also unclear whether additional authors, amended filings or other actions will change the current list of plaintiffs.
There is a related distinction in Anthropic’s position: the company said the pirated datasets were not in the training corpus of commercially released models, while the 2025 court record described portions of the collections being used to train some models. The court record and the company’s settlement representation concern different formulations and scopes. The available account does not establish how those statements reconcile in detail.
The December Hearing Ahead
The next stated milestone in the opt-out cases is a hearing on Anthropic’s motion to dismiss, scheduled for December 17, 2026. The judge’s decision could determine which claims proceed, but the hearing date is not a trial date and does not signal an outcome.
Separately, the settlement calls for payments in installments, including amounts due by September 27, 2027, and requires destruction of the specified pirated collections subject to legal obligations. Further court filings may clarify the claims, the authors involved and the relationship between the 2025 order and Anthropic’s statements about commercially released models.
Key Questions
Did a court rule that training Claude on books was illegal?
No. In June 2025, Judge William Alsup found that training Claude on books was fair use. He separately found that downloading and retaining pirated copies was not fair use.
What did Anthropic agree to pay?
Anthropic agreed to a $1.5 billion settlement covering rights holders of about 500,000 registered books connected to LibGen and PiLiMi. The agreement was finally approved in July 2026.
Did the settlement prove that every pirated book trained Claude?
No. The settlement resolved claims about past conduct without a trial on damages. Anthropic represented that the LibGen and PiLiMi datasets, including portions, were not in the training corpus of commercially released models; the settlement is not a court finding that every listed book was used in such a model.
Are the new author lawsuits decided?
No. Authors who opted out filed new cases in May 2026, and their claims remain pending. A hearing on Anthropic’s motion to dismiss is scheduled for December 17, 2026.
Does Anthropic’s IPO filing say Claude is sentient or could blackmail people?
According to the source account’s review, the S-1 does not make those claims as facts. It discusses model behavior resembling blackmail in testing, debate about potential AI sentience and regulation, and existential concerns raised by others.
Source: ThorstenMeyerAI.com
