🔍 Read the full analysis: Anthropic’s Claude Training Claims: What Investors May Want To Know on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Anthropic’s copyright record combines a 2025 court ruling that model training on books was fair use with a finding that the company illegally downloaded millions of pirated books. A $1.5 billion settlement was approved in July 2026, while new author lawsuits remain unresolved. Anthropic’s IPO filing discusses AI safety risks, but it does not say Claude is sentient or will blackmail people.
Anthropic’s copyright record is drawing investor attention as the AI company prepares to go public: a federal judge ruled in 2025 that training Claude on books was fair use, but also found that the company’s downloading of millions of pirated books was not. A $1.5 billion settlement over those downloads was approved in July 2026, while separate lawsuits by authors who opted out remain unresolved.
The distinction is important because the court did not rule that training Claude on books was illegal. In Bartz v. Anthropic, U.S. District Judge William Alsup’s June 23, 2025 order found that training on books was fair use. He also found that Anthropic’s acquisition and permanent storage of pirated copies was not fair use. The order described more than seven million book copies from Books3, Library Genesis and the Pirate Library Mirror. The ruling addressed a pretrial record; a jury did not decide damages because the case later settled.
The settlement covers claims over approximately 500,000 registered books that Anthropic downloaded from LibGen or PiLiMi, with payments totaling $1.5 billion. That works out to about $3,000 per book before fees, according to the supplied account. The agreement requires Anthropic to destroy the downloaded books and copies, subject to legal-hold obligations. It releases specified claims over past conduct through August 25, 2025, but does not release claims concerning Claude’s outputs.
Anthropic denies wrongdoing. In the settlement, it represented that neither the LibGen nor PiLiMi datasets, nor portions of them, were in the training corpus of any commercially released model. That statement is distinct from the court’s finding about copying and the earlier record concerning training from portions of pirated collections. New lawsuits filed by authors who opted out have not been decided. The supplied source says a hearing on Anthropic’s motion to dismiss is scheduled for December 17, 2026.
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 Exposure in the IPO
For prospective investors, the dispute is not simply a question of whether AI training on copyrighted books is lawful. The court drew a line between using material to train a model and obtaining and retaining pirated copies. That distinction may matter to how investors evaluate Anthropic’s legal exposure, data practices and the costs of settling or litigating similar claims.
The settlement puts a substantial price on one set of past claims, but it does not settle every copyright question involving the company. The agreement’s scope is limited, and the opt-out lawsuits continue. The case also highlights that a payment resolving claims does not amount to a general license for future training on copyrighted works. Investors reading the IPO materials should distinguish settled liabilities from pending claims and broader risks that have not been adjudicated.
The company’s filing also describes risks related to AI behavior and possible regulation. Those disclosures matter as statements of risk, not as evidence that a particular harm has occurred. The source account says the filing refers to model behavior “resembling blackmail” in a staged test, debate over AI sentience and existential concerns raised by others. It does not say Claude is sentient or that the model will blackmail people.
Products Worth Considering
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
As an affiliate, we earn on qualifying purchases.
What the 2025 Ruling Decided
The underlying case, Bartz v. Anthropic, was filed by three authors in August 2024. In his June 2025 order, Judge Alsup treated the company’s activities differently: buying used print books and scanning them was fair use, as was training a model on books; downloading pirated copies to build a permanent library was not. He wrote that buying books later would not erase liability for the earlier copying.
The decision did not resolve every issue through a trial. On the pirated-copy claim, the judge accepted the authors’ version of the facts for the pretrial ruling, while a jury was expected to consider remaining matters, including damages. The parties settled before that could happen. The settlement was finally approved in July 2026, according to the supplied source material.
Separate reporting by The Stanford Daily named writers and books it said were among the affected copies. The source account cautions that it did not independently verify that list against the settlement database. It also says the paper’s description of a May 2026 ruling is inaccurate: the cited ruling was issued in June 2025 in a different case. Those distinctions matter when evaluating claims about specific authors or what a court has decided.
““spectacularly so””
— Judge William Alsup, in the June 23, 2025 order
Pending Claims and Data Questions
The new opt-out authors’ claims have not been ruled on. The scheduled December 17, 2026 hearing concerns Anthropic’s motion to dismiss; it is not a trial or a decision on the merits. The number of plaintiffs, the precise claims that will proceed and any eventual damages remain subject to the court’s actions.
There are also limits to what can be concluded from the settlement record. Anthropic’s representation about commercially released models is not the same thing as a judicial finding on every training dataset or model. The supplied account says the 2025 court record described portions of pirated collections being used to train some models, while Anthropic made a separate representation concerning commercial releases. The full reconciliation of those statements and their precise scope is not established in the source material.
Reports naming particular books and authors should be treated as reported claims unless checked against the settlement database or court records. The source account says it did not verify the Stanford Daily’s list, and it could not confirm Tobias Wolff as a plaintiff in the five related cases it reviewed. Further filings or amendments could change the publicly available record.
The December Hearing and IPO
The next identified court milestone is the December 17, 2026 hearing on Anthropic’s motion to dismiss the opt-out authors’ claims. Any ruling could narrow or advance those cases, but the outcome is not yet known. The settlement payment schedule also extends into 2027, with the source account listing a final $450 million installment due by September 27, 2027.
Investors can compare these developments with Anthropic’s IPO disclosures as the public offering process advances. The filing’s risk language should be read as disclosure of possible legal, regulatory and product risks—not as proof that the risks have materialized. Any additional court filings, settlement updates or changes to the offering documents may clarify the company’s exposure and the status of its claims.
Key Questions
Did a court rule that Claude’s training on books was illegal?
No. Judge William Alsup ruled in June 2025 that training on books was fair use. He separately found that downloading and retaining pirated copies was not fair use.
What does the $1.5 billion settlement cover?
It resolves specified claims related to roughly 500,000 registered books Anthropic downloaded from LibGen or PiLiMi. The agreement does not release claims about Claude’s outputs, according to the supplied account.
Are the newer author lawsuits settled?
No. Authors who opted out filed separate suits, and the 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 will blackmail people?
No, according to the source account. The filing discusses behavior described as “resembling blackmail” in a staged safety test, the debate about AI sentience and existential concerns raised by others. Those are risk disclosures, not claims that Claude is sentient or will blackmail people.
Source: ThorstenMeyerAI.com
