Judge Delays Approval of Anthropic’s $1.5 B AI Copyright Settlement

Anthropic, an AI startup, agreed to pay $1.5 billion to settle claims that it copied copyrighted books to train its language models. The deal would be the largest copyright settlement in U.S. history. On Thursday, U.S. District Judge Araceli Martinez‑Olguin refused to sign off, asking dissenting class members to explain why they oppose the agreement.

Objectors argue that the settlement’s fee structure leaves authors with a “pittance.” Lawyers representing the class seek more than $320 million in fees, while each author would receive roughly $3,000. Pierce Story, an author of two affected works, wrote that every dollar taken by counsel reduces compensation for those harmed. He estimated that lawyer fees translate to $10,000–$12,000 per hour, far above rates in comparable cases such as the T‑Mobile class action, where courts deemed $7,000–$9,500 per hour unreasonable.

The judge’s request for clarification signals concern that the settlement may not fairly balance the interests of plaintiffs and their counsel. Critics say the fee request violates an earlier promise to tie attorney compensation to the amount actually paid to class members. They also note that many eligible authors have not yet registered, reducing the likelihood that they will receive any payment.

If the court ultimately rejects the current terms, Anthropic could be forced to renegotiate a lower fee structure or increase author payouts. A revised settlement would need to address the disparity between legal fees and the modest compensation promised to thousands of writers.

What to watch next: The court’s deadline for additional comments and any subsequent motion by Anthropic’s lawyers will shape whether the settlement proceeds, is revised, or is abandoned entirely.