The US District Court for the Northern District of California has approved a landmark $1.5 billion settlement in the case of Bartz v. Anthropic, setting a new precedent in the ongoing debate over the use of copyrighted material in AI training. The lawsuit alleged that Anthropic had utilized pirated books to train its Claude AI platform, raising significant legal questions regarding copyright infringement and fair use in the realm of artificial intelligence.
This settlement arrives after a protracted legal battle lasting over a year, during which a judge retired and a preliminary approval was granted in September. In early rulings, Judge William Alsup held firm that AI companies are permitted to use legally purchased books for training but drew a clear line against the use of pirated materials. He remarked, “Anthropic had no entitlement to use pirated copies for a central library. Creating a permanent, general-purpose library was not itself a fair use.” This clarification distinguished the purchased, digitized books from pirated ones, asserting that pirated books should have been lawfully acquired.
Judge Araceli Martínez-Olguín recently granted final approval to the settlement, which amounts to $3,000 per book covering over 400,000 books. She dismissed objections to the settlement amount, underscoring the importance of delivering value and closure to impacted authors. As an indication of the settlement’s impact, more than 91% of those affected have claimed their payouts, a noteworthy benchmark of engagement in the class action’s resolution process.
The settlement also includes a substantial allocation for attorneys’ fees—over $100 million—reflecting the complex legal landscape navigated by the legal team. Judge Martínez-Olguín lauded the representation provided, acknowledging the quality and expertise of the attorneys involved. For those seeking further insights into the background and implications of this decision, the announcement is detailed in the original court ruling.
This significant development marks the first settlement among numerous AI-related copyright lawsuits currently weaving through the judiciary. As the largest copyright settlement globally to date, it sets a notable benchmark for the AI industry, encouraging technological companies to exercise greater responsibility in sourcing training data. The decision signals an end to what some have described as the unregulated “Wild West” era of AI training, intending to establish firmer boundaries around the use of copyrighted content. For further context, NPR elaborates on the broader implications of AI training and copyright challenges in recent cases.