A U.S. federal judge has formally approved a landmark $1.5 billion (roughly €1.3 billion) settlement against Anthropic, marking the largest copyright class-action payout in history. The ruling by U.S. District Judge Araceli Martínez-Olguín concludes a closely watched legal battle initiated by authors and publishers who accused the AI developer of illegally downloading and storing millions of pirated books from “shadow libraries” to train its Claude generative AI models.
Under the finalized agreement, affected creators are slated to receive approximately $3,000 for each of the estimated 500,000 copyrighted works used. While the court previously ruled that training AI on legally obtained books could qualify as fair use, it determined that Anthropic’s mass harvesting and storage of unlawfully acquired books constituted severe copyright infringement. With over 91 percent of eligible rightsholders already submitting claims, the record-breaking payout establishes a major legal precedent for how artificial intelligence developers source, license, and manage their training data moving forward.
