The news: A federal judge rejected Anthropic’s agreement to pay at least $1.5 billion to settle a landmark lawsuit brought by a group of authors.

• Judge William Alsup expressed concerns that the ruling would be forced “down the throat of authors,” per Bloomberg Law.
• He added that class members need to be given “very good notice” of the settlement to make sure they can either opt in or opt out, and that lawyers must ensure Anthropic can’t be sued again for the same issue going forward.

In settlement documents, the plaintiffs’ lawyers said it would be the largest publicly reported recovery in US copyright history—an outcome that could open the door to future legal challenges against AI leaders.

The background: Anthropic was accused of scraping a data set of pirated books sourced from Pirate Library Mirror (PiLiMi) and Library Genesis (LibGen), which the company states it obtained through purchase.

In June, a federal judge ruled that Anthropic had the legal right to train models on millions of legally purchased books under the doctrine of “fair use.” However, the judge left room for lawsuits related to the use of pirated books.

Why it matters: Publishers are struggling to contain how AI models access and reuse their content—authorized or not. Arguments against the use of their work in AI summaries and chatbot responses range from reduced web traffic and ad revenues to dilution of the original work’s value.

Meanwhile, AI companies are facing a dwindling supply of online public training data, likely pressuring them to find new sources for future model improvement.

Our take: If approved, the settlement could set a legal precedent for future copyright battles between creators and AI firms. It could also push regulators to be more stringent in requirements for content licensing deals and cause AI companies to move more carefully when scraping data, considering the costs of legal proceedings.