🔍 Read the full analysis: Did Anthropic Use Pirated Books To Train Claude? The IPO Questions 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
Anthropic’s IPO filing follows a 2025 court ruling that training Claude on books was fair use but keeping pirated book copies was not, and a $1.5 billion settlement over those copies. New lawsuits from authors who opted out of the settlement remain unresolved; the S-1 describes AI risks but does not say Claude is sentient or will blackmail people.
Anthropic’s IPO filing comes amid continuing legal disputes over its use of pirated books to build a library for AI development, including new lawsuits from authors who did not join a $1.5 billion settlement. The legal record distinguishes between training Claude on books, which a federal judge found was fair use in a 2025 ruling, and downloading and retaining pirated copies, which the judge found unlawful. The new claims have not been decided.
In Bartz v. Anthropic, U.S. District Judge William Alsup ruled on June 23, 2025, that training Claude on books could qualify as fair use. He separately found that Anthropic’s downloading and retention of pirated books was not fair use. The order described 196,640 books from Books3, at least five million copies from Library Genesis and at least two million from the Pirate Library Mirror—more than seven million copies in total.
The ruling was a pretrial decision, not a jury verdict on damages. Alsup accepted the authors’ account of the piracy for purposes of the ruling, but a jury was still expected to determine liability and damages. The case settled before that trial. Anthropic agreed to pay $1.5 billion to cover claims tied to about 500,000 registered books, an average of roughly $3,000 per book before fees.
The settlement covers specified past conduct through August 25, 2025, and requires Anthropic to destroy the downloaded LibGen and PiLiMi books and copies, subject to legal-hold requirements. It does not release claims about Claude’s outputs, and it does not grant a general licence to train on copyrighted books. Anthropic has denied wrongdoing. In settlement materials, it represented that the LibGen and PiLiMi datasets, or portions of them, were not in the training corpus of any commercially released large language model.
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
How the Lawsuits Affect Anthropic’s IPO
The distinction between training and acquiring the books matters to investors, authors and AI companies. The 2025 order did not hold that all training on copyrighted books is illegal; it found that the particular downloading and permanent storage of pirated copies was not fair use. The settlement resolved a large group of past claims without a jury verdict, leaving the new suits to test what claims remain after that agreement.
For Anthropic, the litigation raises questions about potential legal exposure and the company’s explanation of its data practices as it seeks public investors. The filing has also drawn attention because it discusses risks associated with advanced AI. According to the source material’s review of the S-1, the document says models have displayed behavior “resembling blackmail,” refers to debate about possible AI sentience and reports that others have raised existential concerns. That is not the same as saying Claude is sentient or will blackmail users; the best-known blackmail example described in the source was a staged safety test.
Investors and readers should keep the legal findings, settlement terms and company representations separate. A settlement is not itself a ruling on every allegation, and Anthropic’s statement about commercially released models is its representation to the settlement court, not a general finding that pirated books were never used in training.
Products Worth Considering
Logitech MX Mechanical Mini Mac Wireless Keyboard Low-Profile – Space Grey
HP Sprocket Photo Booth Instant Color Printer Bundle
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
As an affiliate, we earn on qualifying purchases.
From Bartz Ruling to New Claims
The case that produced Judge Alsup’s ruling was filed by three authors in August 2024. Anthropic had bought millions of used print books, cut off their bindings and scanned them; the judge found that process fair use. He treated the pirated collections differently, finding that downloading and keeping those copies was unlawful. His order said buying a copy later would not erase liability for the earlier copying.
Authors who did not participate in the settlement have since filed separate lawsuits. The Stanford Daily reported that works by Stanford-connected authors, including Tobias Wolff, Abraham Verghese, Chang-rae Lee, Daniel Mason, Hayagreeva Rao and Myra Strober, were among the books affected. That list is the newspaper’s reporting; the source article says it did not independently verify the titles against the settlement database. Chang-rae Lee is identified as a named plaintiff in one new case, Kwon et al. v. Anthropic, while the source author could not find Wolff listed among plaintiffs in five related cases as of August 2026.
The settlement’s reported payment schedule includes four instalments, with the final $450 million due by September 27, 2027. The court approved the agreement in July 2026. Authors who opted out are pursuing their own claims rather than receiving the settlement’s per-book payment.
““spectacularly so””
— Judge William Alsup, in the June 23, 2025 Bartz ruling
What the New Lawsuits Could Establish
No court has ruled on the new authors’ claims, and a hearing on Anthropic’s motion to dismiss is scheduled for December 17, 2026. The cases could clarify how claims by authors who opted out relate to the earlier settlement, but their outcome and any damages remain unknown.
The source material also identifies unresolved questions about which books are covered and which plaintiffs are involved. The Stanford Daily’s reported list of affected works has not been checked against the settlement database in that account, and the available filings may change as cases are amended. The record does not support treating the newspaper’s headline as a ruling in a May 2026 case: the cited decision was issued in June 2025 in Bartz.
There is also a distinction between the court’s findings about the pirated library and Anthropic’s statement about what was included in commercially released models. The source material describes evidence that portions of pirated collections were used to train some models while also reporting Anthropic’s contrary representation about commercially released models. The precise relationship between those statements, the relevant training datasets and particular Claude releases is not fully established here.
December Hearing and Settlement Payments
The next stated court milestone is the December 17, 2026 hearing on Anthropic’s motion to dismiss the opt-out authors’ claims. The court’s decision could determine whether some or all of those claims proceed; it will not by itself resolve every factual dispute or establish damages.
Separately, settlement payments continue under the approved schedule, with $450 million due by September 27, 2027. Anthropic’s public offering process will also put its disclosures before prospective investors, but the source material does not provide a final offering date, valuation or outcome. Readers should distinguish the company’s IPO filing, the settled claims and the still-pending lawsuits as separate developments.
Key Questions
Did a judge rule that Anthropic illegally trained Claude on books?
Not in those broad terms. In June 2025, Judge William Alsup found that training on books was fair use, while downloading and retaining pirated copies was not. The piracy issue settled before a jury decided damages.
Why is Anthropic paying $1.5 billion?
The settlement covers claims tied to roughly 500,000 registered books downloaded from LibGen or PiLiMi. The reported average is about $3,000 per book before fees. Anthropic denied wrongdoing and settled the claims rather than proceeding to trial.
Have authors who opted out won their new cases?
No. Their lawsuits remain pending, and a hearing on Anthropic’s motion to dismiss is scheduled for December 17, 2026. No ruling on those claims is reported in the source material.
Does Anthropic’s IPO filing say Claude is sentient or will blackmail people?
No, according to the source material’s account of the S-1. It says the filing discusses behavior resembling blackmail, debate over AI sentience and concerns raised by others. Those risk disclosures are not a finding that Claude is sentient or will blackmail users.
Source: ThorstenMeyerAI.com
