🔍 Read the full analysis: What The Anthropic Allegations Say About Claude’s Training Data on ThorstenMeyerAI.com
Get smart everyday buys 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 2025 federal court order distinguished between Anthropic’s use of books to train Claude, which Judge William Alsup found was fair use, and the company’s downloading and retention of pirated book collections, which he found was not. Anthropic later agreed to a $1.5 billion settlement over claims tied to those copies, while new lawsuits filed by authors who opted out remain unresolved. The claims are not a final ruling that Claude’s commercially released models were trained on the pirated collections.
Anthropic’s book-training dispute is back in focus after The Stanford Daily reported that works by Stanford-affiliated authors were among books allegedly copied without permission. The court record is more specific than the headline: in June 2025, a judge found that training Claude on books was fair use, but that Anthropic’s downloading and permanent retention of pirated book copies was not. New lawsuits filed by authors who did not join a settlement have not yet been decided.
In Bartz v. Anthropic, U.S. District Judge William Alsup ruled on June 23, 2025, that using books to train the company’s models was fair use under copyright law. He also found that Anthropic’s purchase of print books and scanning them for training qualified as fair use. The ruling drew a different conclusion about copies downloaded from Books3, Library Genesis (LibGen) and the Pirate Library Mirror (PiLiMi) and kept in a permanent library.
The order described about 196,640 Books3 titles, at least five million copies from LibGen and at least two million from PiLiMi. It found that downloading and retaining those copies was not fair use. The judge accepted the authors’ account of the piracy for purposes of the pretrial ruling; a jury did not decide the remaining issues because the case later settled.
Anthropic agreed to a $1.5 billion settlement, which received final approval in July 2026, according to the supplied source. The settlement covers claims over roughly 500,000 registered books and provides about $3,000 per book before fees. Anthropic denied wrongdoing and represented to the settlement court that the LibGen and PiLiMi datasets, or portions of them, were not in the training corpus of any commercially released model. The agreement also requires destruction of the downloaded copies, subject to legal-hold obligations, and does not release claims about model outputs.
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 Ruling Means for Claude
The distinction matters because copyright infringement allegations do not equal a finding that model training itself was unlawful. Alsup’s ruling addressed separate conduct: learning from books for model training, scanning purchased books, and downloading pirated copies to retain. Treating those as one issue can give readers a misleading account of what the court decided.
The case also leaves practical questions for authors and AI developers. The settlement compensates eligible rights holders for covered books and resolves specified past claims, but it is not a general licence for Anthropic to train on any copyrighted work. Nor does the settlement amount establish that each book affected a model’s outputs or commercial performance. The claims about new plaintiffs remain unsettled, so the next court decisions may clarify how those disputes proceed.
Products Worth Considering
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
Dell XPS 13 9350 Laptop, 13.4 FHD+ Display, Intel Ultra 7 256V, 16GB DDR5, 1TB SSD
Dell UltraSharp U2725QE 27-Inch 4K UHD IPS Black Monitor
As an affiliate, we earn on qualifying purchases.
The original case was brought by three authors in August 2024. In June 2025, Alsup concluded that training on books could be transformative fair use, describing Claude’s training as an effort “to turn a hard corner and create something different.” He separately said that acquiring pirated copies for a permanent library could not be excused by later purchasing books. The case settled before a jury considered damages.
The Stanford Daily reported that books by Tobias Wolff, Abraham Verghese, Chang-rae Lee, Daniel Mason, Hayagreeva Rao and Myra Strober were among the copies at issue, and said ten Wolff works were affected. Those details should be treated as the newspaper’s reporting: the supplied account says the book list was not independently checked against the settlement database. It also says Chang-rae Lee appears as a plaintiff in one new suit, while Wolff was not found among plaintiffs listed in five related cases as of August 2026.
Authors who opted out of the settlement filed new cases in May 2026. Those filings are allegations, not findings by a judge. The supplied source says Anthropic has argued it will not settle above the value of the Bartz claim, and that one author accepted $2,000 plus costs. Neither point determines the merits or outcome of the other cases.
“to turn a hard corner and create something different”
— Judge William Alsup, in the June 23, 2025 order
Open Questions in the Lawsuits
No ruling has been issued on the May 2026 opt-out claims, and the supplied source says a hearing on Anthropic’s motion to dismiss is scheduled for December 17, 2026. The allegations in those cases therefore remain unproven. It is also unclear from the available material which authors and titles are included in each action and how the claims will be affected by the settlement’s limits.
There is a separate distinction between the court record and Anthropic’s statement about commercial models. The 2025 record described pirated collections and their use in training some models, while Anthropic represented in the settlement that those collections, or portions of them, were not in the training corpus of commercially released models. The supplied account does not explain the full dataset and model timeline needed to reconcile those descriptions. The Stanford Daily’s reported title list also has not been verified here against the settlement database.
December Hearing and Case Filings
The next listed milestone is the December 17, 2026 hearing on Anthropic’s motion to dismiss the opt-out authors’ claims. A hearing is not a decision on the underlying allegations; the court’s eventual ruling will determine whether some or all claims can proceed. Further filings may identify the plaintiffs, works and legal arguments more clearly.
The settlement provides for payments in instalments, with the supplied account listing a final $450 million payment by September 27, 2027. Readers should distinguish that administration of the settlement from the separate lawsuits, which concern authors who opted out. Until the court rules, the new claims remain allegations and do not change what Alsup decided in the 2025 case.
Key Questions
Did a judge rule that Claude’s training 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 keeping pirated copies was not fair use.
What does the $1.5 billion settlement cover?
According to the supplied source, the settlement covers claims involving roughly 500,000 registered books downloaded from LibGen or PiLiMi, at about $3,000 per book before fees. It addresses specified past conduct, not every possible copyright claim or claim about Claude’s outputs.
Has a court ruled on the new author lawsuits?
No ruling is reported. Authors who opted out filed new suits in May 2026. Anthropic’s motion to dismiss is scheduled for a December 17, 2026 hearing, and the allegations remain unresolved.
Did Anthropic say commercially released Claude models used the pirated books?
No. Anthropic represented in the settlement proceedings that the LibGen and PiLiMi datasets, or portions of them, were not in the training corpus of any commercially released model. That is Anthropic’s stated position; the supplied material does not provide enough detail to resolve the relationship between that statement and the broader descriptions in the court record.
Source: ThorstenMeyerAI.com
