🔍 Read the full analysis: A Closer Look At Claims Claude Was Trained On Pirated Books on ThorstenMeyerAI.com
Get hardware and tech essentials 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 federal judge ruled in June 2025 that Anthropic’s use of books to train Claude was fair use, while finding that the company’s downloading and permanent storage of pirated books was not. Anthropic later agreed to a $1.5 billion settlement covering about 500,000 books, and authors who opted out have filed separate lawsuits that have not been decided.
A 2025 federal court ruling found that Anthropic’s use of books to train Claude was fair use, but ruled that downloading and keeping millions of pirated books was not. That distinction matters as the company faces a $1.5 billion settlement over roughly 500,000 books and separate lawsuits from authors who opted out; those newer claims have not yet been decided.
In Bartz v. Anthropic, Judge William Alsup’s June 23, 2025 order separated the company’s training practices from how it obtained and stored some source books. He concluded that using books to train a model was fair use and that Anthropic’s scanning of print books it had bought was also fair use. But he found that downloading pirated copies and retaining them in a permanent library was not. The order described more than seven million copies from collections including Books3, Library Genesis and the Pirate Library Mirror.
The ruling was a pretrial summary-judgment order, not a jury verdict on damages. On the pirated-copy issue, the judge accepted the authors’ version of the facts for purposes of the decision; a jury was expected to consider remaining issues before the case settled. The downloads were not the same legal question as whether model training itself was fair use.
Anthropic agreed to pay $1.5 billion to resolve claims for owners of about 500,000 registered books downloaded from Library Genesis or Pirate Library Mirror. The settlement provides roughly $3,000 per book before fees and requires the company to destroy the downloaded collections and copies, subject to legal-hold obligations. The agreement covers past conduct through August 25, 2025, and does not release claims concerning Claude’s outputs. Anthropic has denied wrongdoing.
In settlement materials, Anthropic represented that neither the Library Genesis nor Pirate Library Mirror datasets, nor portions of them, were in the training corpus of any commercially released large language model. That statement should be kept distinct from the court’s finding about downloading and storing the books. The source account reports both that portions of pirated collections were used to train some Anthropic models and that Anthropic made this narrower representation concerning commercially released models; the precise scope and reconciliation of those statements are not established here.
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
Training and Piracy Are Separate Issues
The distinction affects how readers should understand claims that Claude was “trained on pirated books.” The court record described pirated books being downloaded and retained, and Judge Alsup ruled that this conduct was not fair use. But the same ruling found the training use of books fair use. Treating the decision as a finding that Claude’s training was illegal would overstate what the judge decided.
The settlement resolves claims for the covered books without turning the agreement into a general license for training on copyrighted material. Its terms address specified past conduct and registered works, while leaving claims about model outputs outside the release. For authors and publishers, the case is a major financial resolution; for AI companies, it also highlights that the means of acquiring and storing training material can raise legal questions distinct from the use made of that material.
The separate opt-out cases may test issues that the settlement did not resolve. Their outcome could affect how copyright claims involving AI training proceed, but no ruling on those allegations is in place. Readers should also treat allegations in new complaints as claims, not findings of liability.
Products Worth Considering
Apple MacBook Pro 14‑inch (2021) with M1 Pro, 16 GB RAM, 512 GB SSD – Space Gray
Logitech MX Mechanical Mini Mac Wireless Keyboard Low-Profile – Space Grey
Logitech MX Keypad Programmable Macro Pad, Graphite
As an affiliate, we earn on qualifying purchases.
Bartz v. Anthropic was filed by three authors in August 2024. In the June 2025 order, Alsup described Anthropic’s purchase and scanning of millions of used print books as fair use, while distinguishing those books from pirated digital copies kept in a library. He wrote that buying a copy later would not erase liability for the earlier theft of a pirated copy. The jury trial on remaining matters did not take place after the parties reached a settlement.
The settlement was finally approved in July 2026, according to the source account. It sets payments in four instalments: $300 million in October 2025, $300 million after final approval, $450 million by September 25, 2026, and $450 million by September 27, 2027. The source also reports that a court reduced the lawyers’ fee to about 6.8 percent, citing the author-advocacy site Writer Beware.
Separately, authors who opted out filed new suits in 2026. One case, Kwon et al. v. Anthropic, names author Chang-rae Lee, according to the source’s review of filings. The Stanford Daily reported that works by several Stanford-connected writers were among pirated copies, including Tobias Wolff, Abraham Verghese, Daniel Mason, Hayagreeva Rao and Myra Strober. The source author says the book list was not independently checked against the settlement database, so those details remain attributed to the paper.
““spectacularly so””
— Judge William Alsup, in the June 23, 2025 order
Open Questions in Opt-Out Cases
No judge has ruled on the claims in the new opt-out lawsuits, and the source account says a hearing on Anthropic’s motion to dismiss is scheduled for December 17, 2026. The allegations in those cases should not be described as established findings. The record summarized here also does not establish the final outcome of that hearing.
Some details reported about which authors’ books were included remain unverified in the source account. Its author says the Stanford Daily’s book list was not checked against the settlement database and could not find Tobias Wolff listed among plaintiffs in five related cases as of August 2026; Wolff could be involved in another action or a later amendment. The settlement covers registered books in the specified pirated collections, not every work attributed to an author in media reports.
The relationship between the court record’s discussion of pirated collections and Anthropic’s statement about commercially released models also requires precise wording. The available material does not fully explain which particular datasets or model versions were involved in each training stage. Nor does the settlement itself resolve whether any output from Claude infringes copyright.
December Hearing and Settlement Payments
The next identified court milestone is the December 17, 2026 hearing on Anthropic’s motion to dismiss the opt-out authors’ claims. The court’s action after that hearing will indicate whether some or all of those claims proceed; it will not, by itself, establish that the allegations are true.
The approved settlement also has payments scheduled through September 27, 2027, including the final $450 million instalment. Further reporting should distinguish payment administration and any later court rulings from the separate question of how particular books were used to train particular models. Until the new cases produce decisions, the 2025 ruling and settlement remain the clearest confirmed record of the claims covered in this dispute.
Key Questions
Did a judge rule that Claude’s training on books was illegal?
No. In June 2025, Judge William Alsup found that using books to train Claude was fair use. He separately ruled that downloading and permanently keeping pirated copies was not fair use.
What does the $1.5 billion settlement cover?
It covers claims involving about 500,000 registered books downloaded from Library Genesis or Pirate Library Mirror, with payments averaging roughly $3,000 per book before fees. The agreement addresses past conduct through August 25, 2025, and does not release claims about Claude’s outputs.
Have the authors who opted out won their lawsuits?
No ruling on those new lawsuits is reported. Anthropic’s motion to dismiss is scheduled for a hearing on December 17, 2026; the allegations remain unresolved.
Did Anthropic say pirated books were used to train Claude?
Anthropic denied wrongdoing and represented in settlement materials that the LibGen and PiLiMi datasets, or portions of them, were not in the training corpus of any commercially released model. That statement should be reported alongside, not conflated with, the court’s findings about downloading and storing pirated copies. The available account does not fully reconcile the scope of the claims about training different models.
Source: ThorstenMeyerAI.com
