🔍 Read the full analysis: Anthropic’s IPO Spotlight: Claims About Claude’s Training Books on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Anthropic’s IPO filing has renewed attention to copyright claims about books used in Claude’s development. A 2025 ruling found model training on books to be fair use in the case before it, but said Anthropic’s downloading of millions of pirated copies was not; a $1.5 billion settlement resolved that case without a trial on damages. New author lawsuits remain unresolved.
Anthropic’s filing to go public has renewed scrutiny of claims that books were used to train Claude, but the legal record distinguishes between training on books and downloading pirated copies. In a June 2025 ruling, a federal judge found that training Claude on books was fair use in the case before him, while finding that Anthropic’s copying of millions of pirated books was not; a later $1.5 billion settlement resolved that case, and separate authors’ lawsuits remain pending.
The dispute stems in part from Bartz v. Anthropic, filed by three authors in August 2024. On June 23, 2025, U.S. District Judge William Alsup ruled that Anthropic’s use of books to train its model was transformative fair use. He also found that scanning books Anthropic had purchased was fair use. But the judge drew a different line at the company’s downloading and retention of pirated books from online collections.
The ruling described more than seven million copies, including books from Books3, Library Genesis and the Pirate Library Mirror. Alsup found the copying of those pirated collections unlawful on the facts presented in the case. The decision did not determine damages after a trial: the parties settled before a jury could decide that issue. The settlement, finally approved in July 2026, totals $1.5 billion and covers claims tied to roughly 500,000 registered books. It requires destruction of the downloaded collections, subject to legal-hold obligations.
Anthropic has disputed wrongdoing. In settlement materials, the company represented that the LibGen and PiLiMi datasets, or portions of them, were not included in the training corpus of any commercially released model. That statement does not erase the earlier court record: the 2025 order addressed copying and training evidence in the Bartz case, while the settlement resolved claims without a trial on damages. The settlement releases certain past claims through August 25, 2025; it does not grant a general license to use books or release claims about Claude’s 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
The distinction matters to authors, publishers and AI companies because copyright disputes can turn on how works were obtained as well as how they were used. The ruling in Bartz did not establish that every use of copyrighted books to train an AI system is lawful. It decided particular questions on the record before the court: training and scanning purchased books were fair use, while downloading and keeping pirated copies was not.
The settlement also puts a substantial financial figure on one case without resolving the broader debate over AI training. Authors whose books were part of the settlement may qualify for compensation under its terms, but the deal does not settle separate authors’ allegations or determine whether Claude’s generated outputs infringe copyright. For readers, the key point is that the $1.5 billion agreement is not a judgment that all Claude training was illegal.
The IPO filing has added investor and public attention to these issues. The source account says Anthropic’s S-1 discusses model behavior that can resemble blackmail in testing, possible regulation arising from public debate about AI sentience, and concerns raised by others about existential risks. Those descriptions should not be recast as a claim that the filing says Claude is sentient or will blackmail people; the filing language, as described in the source, presents risks and debate rather than such conclusions.
Products Worth Considering
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
Amazon Kindle Paperwhite (12th Generation, 2024), 16 GB, 7-inch Display, Jade
DJI Osmo Pocket 3 Filter Bundle – Vlogging Camera with 1 CMOS Sensor and 4K/120fps
As an affiliate, we earn on qualifying purchases.
How the Copyright Cases Differ
The Stanford Daily reported that books by Stanford-affiliated writers, including Tobias Wolff, Abraham Verghese, Chang-rae Lee and Daniel Mason, appeared among pirated copies linked to the dispute. It also reported that ten of Wolff’s works were affected. Those details are the student newspaper’s reporting; the source account says the book list was not independently checked against the settlement database. Wolff told the paper that using his work without payment “is a form of slave labor.” That is his characterization, not a court finding.
The source account also flags a timeline issue in the newspaper’s framing: the relevant ruling was issued in June 2025 in Bartz, not in a May 2026 lawsuit. In May 2026, according to that account, authors filed new suits. Chang-rae Lee is listed as a named plaintiff in one of those actions, Kwon et al. v. Anthropic; the account says it could not find Wolff among the plaintiffs listed in five related cases as of August 2026. The names and claims in the new cases should not be treated as findings by a judge.
The earlier case ended through a settlement, not a final jury verdict on damages. Its payment terms cover eligible registered books in the settlement class, with the source reporting an average of about $3,000 per book before fees. The agreement concerns past conduct and does not resolve all questions about how copyrighted material may be used in future AI training.
““is a form of slave labor””
— Tobias Wolff, as quoted by The Stanford Daily
Claims Still Before the Courts
The new lawsuits have not been decided. The source account reports that a hearing on Anthropic’s motion to dismiss is set for December 17, 2026. Until the court rules, allegations in those cases remain claims, not findings. The account also says one author accepted $2,000 plus costs and that Anthropic’s lawyers said the company would not settle “above the Bartz claim value”; those details do not establish how the pending cases will be resolved.
Some factual questions also remain outside what can be established from the material provided. The reported list of Stanford authors’ books has not been checked against the settlement database, and the account does not establish whether additional plaintiffs or amended filings have changed the lists since August 2026. The settlement’s terms, Anthropic’s statements and the ruling address different parts of the dispute; they do not provide a single court determination about every book, model or output at issue.
The precise scope of the IPO filing’s AI-risk disclosures also depends on its wording. The source describes the filing’s discussion but does not reproduce the full document. The claims circulating online that the S-1 says Claude could blackmail people, become sentient or end humanity therefore should not be treated as quotations or direct conclusions from the filing.
December Hearing and IPO Review
The next reported court milestone is the December 17, 2026 hearing on Anthropic’s motion to dismiss the new authors’ claims. The court’s decision could determine whether some or all of those claims proceed, but it would not by itself settle the underlying copyright questions. Any later ruling or settlement would need to be reported separately from the 2025 Bartz decision.
Investors and the public can also assess the company’s IPO disclosures as the filing process continues. The available source does not provide an IPO timetable or say whether regulators have completed their review, so those details remain open. For now, the established record is narrower than some viral summaries suggest: a judge found training on books to be fair use in one case, found pirated copying unlawful, and the parties settled before a jury determined damages.
Key Questions
Did a judge rule that Claude was illegally trained on books?
Not in those broad terms. In the June 2025 Bartz case, Judge William Alsup found that training Claude on books was fair use on the record before him, while ruling that downloading and retaining pirated copies was not.
What did Anthropic agree to pay?
The settlement totals $1.5 billion and covers claims involving roughly 500,000 registered books. The parties settled before a jury decided damages, and the agreement applies to specified past claims rather than granting a general training license.
Are the new authors’ lawsuits resolved?
No. The source account says the new cases remain pending and that a hearing on Anthropic’s motion to dismiss is scheduled for December 17, 2026. Allegations in those suits have not been decided by a court.
Does Anthropic’s IPO filing say Claude is sentient or will blackmail people?
The source describes the S-1 as discussing behavior that can resemble blackmail in testing, debate about AI sentience and concerns raised by others about existential risks. That is not the same as the filing asserting that Claude is sentient or will blackmail people.
Source: ThorstenMeyerAI.com
