Anthropic Ordered to Pay $1.5B in Landmark Copyright Settlement Over AI Training Data

Anthropic, the $18.3 billion AI startup, has agreed to a historic $1.5 billion settlement with authors who alleged the company used pirated books to train its large language models. The Northern District of California case sets a precedent for copyright enforcement in generative AI, requiring destruction of disputed datasets.

The settlement compensates authors at approximately $3,000 per book—the largest publicly reported copyright payout to date. Plaintiffs Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson successfully challenged Anthropic's 'fair use' defense by proving the training materials included illegally sourced content.

This ruling sends shockwaves through the AI sector, particularly for firms relying on scraped or unlicensed data. The mandated dataset destruction could force costly retraining efforts across the industry, potentially impacting development timelines for next-generation models.