Anthropic Settles Landmark AI Copyright Class Action for $1.5 Billion
In a historic resolution to one of the most high-profile artificial intelligence copyright disputes, Anthropic PBC agreed to a $1.5 billion settlement in the class-action lawsuit Bartz et al. v. Anthropic PBC (Case No. 3:24-cv-05417) in the U.S. District Court for the Northern District of California. The settlement, which was reached in late August 2025 and approved by the court in late September 2025, represents one of the largest copyright litigation settlements in history and sets a massive precedent for the legal liabilities of generative AI developers.
The lawsuit, originally filed in August 2024 by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, accused Anthropic of illegally downloading and utilizing pirated datasets—specifically Library Genesis (LibGen) and Pirate Library Mirror (PiLilMi)—to train its Claude line of large language models. These datasets contained nearly 500,000 copyrighted books with International Standard Book Numbers (ISBNs) or Amazon Standard Identification Numbers (ASINs) registered with the U.S. Copyright Office.
Key Terms of the Settlement
The settlement establishes a structured process for compensating affected authors and publishers while enforcing strict operational changes on Anthropic:
- Financial Compensation: Anthropic faces a total settlement fund of $1.5 billion to pay out valid claims submitted by legal or beneficial owners of the pirated works on the covered list.
- Data Destruction: Anthropic is required to destroy all copies of the books downloaded from the LibGen and PiLilMi datasets.
- No Future Licensing Rights: Crucially, participating in the settlement does not grant Anthropic a license to use these copyrighted works in the future. The release of claims is strictly retroactive, covering past training and development activities up to August 26, 2025. It does not release claims for any future infringements or infringing outputs generated by Claude.
- Timeline: The deadline for class members to exclude themselves (opt out) was January 7, 2026, and the deadline to submit a claim form is March 23, 2026. A final settlement fairness hearing was scheduled for April 23, 2026.
Regulatory and Market Implications
By settling for such a massive sum, Anthropic has signaled to the broader AI industry that relying on pirated datasets carries catastrophic financial and legal risks. The settlement demonstrates that copyright licensing is a viable, necessary path forward for AI development, rather than an existential threat. It incentivizes AI developers to transition toward legitimate, permission-based licensing agreements with authors and publishers for training data, rather than relying on unauthorized scraping or pirated repositories.