Introduction to the Case
A significant settlement has been proposed in the case of Bartz v. Anthropic, with Anthropic agreeing to pay $1.5 billion. This settlement is related to claims that the company downloaded pirated books to train its AI model, Claude. If approved, this would be the largest copyright recovery in the US to date. A preliminary approval hearing was scheduled, marking a crucial step in the process.
Background of the Case
In June, Judge William Alsup ruled that training AI on lawfully obtained books can be considered fair use, but copying and storing millions of pirated books constitutes infringement. This ruling paved the way for settlement discussions between the parties involved. The plaintiffs alleged that Anthropic had pulled at least 7 million copies of books from piracy sites, including Library Genesis and Pirate Library Mirror.
Settlement Details
The proposed settlement would pay approximately $3,000 per eligible title, with an estimated class size of around 500,000 books. This agreement is significant, with Justin Nelson, counsel for the authors, stating that it is likely the largest copyright recovery ever. The payout would be made in four tranches after court approvals, with $300 million paid soon after preliminary approval, followed by additional payments over the next two years, including interest accrued in escrow.
How Payouts Would Work
The fund would be distributed according to a specific schedule, with the first payment of $300 million made after preliminary approval, followed by $300 million after final approval, and then $450 million at 12 months and $450 million at 24 months. A final "Works List" is due on October 10, which will be used to create a searchable database for claimants. The agreement also requires the destruction of pirated copies and only resolves past conduct.
Impact and Implications
This settlement has important implications for the use of AI tools in content workflows. It highlights the importance of provenance and is likely to lead to more licensing deals and clearer disclosures from vendors about their training data sources. For publishers and creators, the per-work payout sets a reference point that may strengthen their negotiating leverage in future licensing talks.
Looking Ahead
The judge’s consideration of preliminary approval marks a significant step forward in the process. If granted, the notice process would begin in the fall, followed by payments to rightsholders after final approval and claims processing. This settlement serves as a landmark case, emphasizing the need for copyright respect and proper licensing in the development and training of AI models.
Conclusion
The proposed $1.5 billion settlement in Bartz v. Anthropic underscores the importance of respecting copyrights and the potential consequences of using pirated materials for training AI. As the use of AI becomes more prevalent, cases like this will continue to shape the landscape of content creation, licensing, and the responsible use of technology. The outcome of this case will have far-reaching implications for authors, publishers, and AI developers, emphasizing the need for transparency, fairness, and adherence to copyright laws.