🔍 Read the full analysis: Anthropic’s AI Training Controversy: Pirated Books And IPO Stakes 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
Anthropic’s IPO filing has drawn attention to a separate copyright dispute over pirated books. A 2025 court ruling said training on books was fair use but found that keeping more than seven million pirated copies was not; a $1.5 billion settlement was approved in July 2026, while new claims by authors who opted out remain unresolved.
As Anthropic moves toward a public offering, its copyright record is drawing renewed scrutiny: a June 2025 federal ruling found that training Claude on books was fair use, but that downloading and retaining more than seven million pirated book copies was not. A $1.5 billion settlement resolving claims by a group of authors was approved in July 2026, while separate claims by authors who opted out have not been decided.
The underlying case, Bartz v. Anthropic, was brought by three authors in August 2024. In a June 23, 2025 summary-judgment order, U.S. District Judge William Alsup distinguished between training models on books and acquiring copies to create a permanent digital library. He ruled that training Claude on books was fair use and also found that Anthropic’s purchase and scanning of print books qualified as fair use. He rejected the use of pirated copies as a defense, writing that later buying a copy would not absolve the company of liability for what he called theft.
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. That finding was made at summary judgment, before trial; the order accepted the authors’ version of the facts on the pirated-copy issue, and a jury would have been asked to decide remaining matters, including damages. The case settled instead. The settlement, approved in July 2026, provides $1.5 billion for roughly 500,000 registered books downloaded from LibGen or PiLiMi, or about $3,000 per book before fees. It also requires destruction of those downloaded books and their copies, subject to legal-hold obligations.
Anthropic has denied wrongdoing. In the settlement process, the company represented that the LibGen and PiLiMi datasets, and portions of them, were not in the training corpus of any commercially released large language model. The source account notes tension between that representation and the 2025 court record, which said portions of the pirated collections were used to train some Anthropic models. The company’s statement about released models is narrower than a claim that no pirated material was ever used in training. The settlement releases claims concerning past conduct through August 25, 2025, but explicitly does not 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
Copyright Claims Follow Anthropic to IPO
The dispute matters to investors and authors because it concerns both the cost of building AI systems and the legal basis for using copyrighted material in training. The $1.5 billion settlement resolves one major set of claims, but it is not a court ruling that settles every question about AI training or copyright. Its scope is limited to participating authors and past conduct described in the agreement.
The financial figure also should not be treated as a final measure of Anthropic’s liability across all books or all AI-related claims. Authors who opted out filed new cases in May 2026, and their allegations remain unresolved. The settlement requires destruction of specified pirated copies; it does not grant a general licence to use books for training. That distinction may matter to how companies document training data and manage copyright exposure as they seek investment or public-market scrutiny.
Anthropic’s filing has also been discussed alongside online claims that it warns Claude could blackmail people or become sentient. According to the source account’s review of the S-1, the filing describes model behavior “resembling blackmail,” including a well-known staged safety test, and discusses debate over AI sentience and existential risks. Those disclosures are risk statements, not findings that Claude is sentient or that it has blackmailed people.
Products Worth Considering
Apple iPad Air (2025) 13‑inch Wi‑Fi 256 GB – Space Gray (M3 chip)
Bose S1 Pro+ Portable PA Speaker
Jackery Solar Generator 1000 v2 with 200W Solar Panel
As an affiliate, we earn on qualifying purchases.
The Ruling, Settlement and New Suits
The legal timeline is central to interpreting the headlines. The Stanford Daily reported that books by Stanford-affiliated authors, including Tobias Wolff, Abraham Verghese, Chang-rae Lee, Daniel Mason, Hayagreeva Rao and Myra Strober, were among the pirated copies. It reported that ten of Wolff’s works were affected. Those book-list details are the newspaper’s reporting; the source account says it did not independently verify them against the settlement database.
The source account also says the student paper linked the piracy finding to a May 2026 lawsuit, although the ruling it identified was the June 2025 order in Bartz, a different case. The May 2026 development was that authors who opted out filed new lawsuits, not that a judge issued a new ruling against Anthropic. Chang-rae Lee is listed as a plaintiff in one of the new actions, Kwon et al. v. Anthropic; the source account could not find Wolff among the plaintiffs listed in five related cases as of August 2026.
Settlement terms cover registered books copied from LibGen or PiLiMi, with payments scheduled in instalments. The agreement concerns past conduct through August 25, 2025, and does not settle claims about model outputs. Anthropic’s S-1 is a separate document: it describes business risks for prospective investors, rather than deciding the copyright cases or establishing that every risk described has occurred.
““Buying a copy later will not absolve it of liability for the theft.””
— Judge William Alsup, as quoted in the court order
Claims Still Awaiting a Court
No ruling has been issued on the claims brought by the authors who opted out of the settlement. The source account says Anthropic sought dismissal and that a hearing is scheduled for December 17, 2026; the outcome and any later trial schedule are not yet known. The allegations in those suits should not be presented as findings of liability.
There is also a dispute in how the record is summarized. The 2025 order found pirated copies were downloaded and described portions of those collections as used in training, while Anthropic represented in the settlement that the collections or portions of them were not in the training corpus of any commercially released model. The distinction between use in training and inclusion in released-model training corpora is important, and the supplied material does not fully reconcile the statements. It is also not clear from that material whether Wolff is a party to another action or a later amendment.
December Hearing and IPO Disclosure
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 which claims proceed, but no outcome can be assumed ahead of the hearing. Settlement payments and the required destruction of specified copies remain governed by the agreement and its schedule.
Anthropic’s reported IPO filing puts its legal and operational risks before prospective investors, but the source material does not establish the offering’s timetable, terms or final regulatory review. Further filings, court decisions and clarifications from the company may add detail. Until then, the established record remains a 2025 fair-use ruling with a distinct finding against retaining pirated copies, a settled class of claims, and separate lawsuits still in progress.
Key Questions
Did a court rule in 2026 that Anthropic illegally trained Claude on books?
No. The cited ruling was issued in June 2025 in Bartz v. Anthropic. Judge Alsup found that training on books was fair use but rejected the downloading and retention of pirated copies. The May 2026 development was the filing of new lawsuits by authors who opted out, not a new ruling.
What does the $1.5 billion settlement cover?
The agreement covers claims over roughly 500,000 registered books downloaded from LibGen or PiLiMi, at about $3,000 per book before fees. It addresses past conduct through August 25, 2025, and does not release claims about Claude’s outputs.
Has Anthropic admitted that Claude was trained on pirated books?
Anthropic denied wrongdoing and represented in the settlement process that the LibGen and PiLiMi datasets, or portions of them, were not in the training corpus of any commercially released model. The 2025 court record separately described portions of the collections as used to train some Anthropic models. Those statements concern different scopes, and the supplied material does not fully resolve the apparent tension.
What happens to the authors who opted out of the settlement?
Their claims are being pursued in separate lawsuits and have not been decided. A hearing on Anthropic’s motion to dismiss is scheduled for December 17, 2026, according to the source material.
Source: ThorstenMeyerAI.com
