Anthropic AI to Pay $1.5 Billion for Pirated Books and Creative Content.
Check Below if Your Work is Included in the Settlement.
By: Ilya Ilienko, CMA, CPA, MBA
Why Does Anthropic Settle?
Anthropic AI agrees to pay $1.5 Billion to settle a copyright class-action lawsuit (Bloomberg Law).
Thursday Sept. 25, 2025, Judge Alsup of the California Northern District allowed a preliminary approval of a settlement in the class action lawsuit brought by three book authors VS. Anthropic (NPR). The authors’ copywritten books were among the several million downloaded by Anthropic to help train its Claude AI models.
Settlement Terms and Judicial Concerns
Anthropic agreed to pay approximately $3,000 for each of the ~half-million works the authors demonstrated were found in pirate libraries of millions of books (Authors Alliance). Judge Alsup said Anthropic “refused to come clean” with detail regarding which pirated works were used to train its AI model and why the company decided to maintain them in a “research library” of its own. In total Anthropic downloaded more than 7 million works from “shadow libraries”.
The Judge further stated that the settlement is “nowhere close to complete” and expressed concern that either or both sides’ lawyers maybe reaching a deal behind closed doors that will be shoved “down the throat of authors.” Anthropic in its own right said it faced “inordinate pressure” to work out a deal in an effort to close the door on upwards of $1 Trillion in statutory damages it was facing at trial.
Huge Implications for the AI Sector
This litigation settlement for copyright infringement is one of the largest in US history — and among the most complex (OpenClassActions). Content creators and owners of creative content place blame on AI platforms of stealing all sorts of material, including: books, movies, songs, news articles, academia research and then feeding all that information into its LLM models to train its AI platforms — which in turn increases the value of the AI platform. Content creators get nothing, and the AI platform increases its worth on the writers’ dime.
OpenAI, Meta, and Midjourney are likewise dealing with problems of their own dealing with dozens of cases from content creators, some of whom are as large as the New York Times — regarding infringement and trademark dilution. While music publishers have filed a separate suit against Anthropic over its alleged wholesale copying of lyrics.
The next important milestone in this class-action is October 10, 2025 — the submission of an updated Works List and proposed Plan of Allocation (Anthropic Settlement Portal).
Effect on Other AI Platforms
This class-action sets a precedent: authors have enforceable rights against OpenAI, Meta, and other platforms, because the companies admitted to acquiring and storing material from the same illegal internet sources.
Mind you, Judge Alsup approved using properly acquired books to train AI models — the legal issue concerns pirated content. One major challenge is identifying damages appropriately, since blanket sums do not reflect the diversity of works, ranging from articles to books, academia to fiction.
Search and Submit Your Work
Submit a Claim Form by March 23, 2026 to join the class-action and be eligible to receive a payment (Authors Guild).
Search for your or other eligible works here (Works Lookup).