Judge approves a $1.5B Anthropic settlement over pirated books used to train the Claude chatbot

A federal judge in San Francisco has approved a 1.5 billion dollar settlement between AI company Anthropic and thousands of authors. The court ruling mandates compensation of approximately 3,000 dollars per book for authors whose copyrighted works were included in the training data for the Claude chatbot.

This legal decision concludes a high-profile class-action lawsuit initiated in 2024 by thriller novelist Andrea Bartz and others. The judge noted that the agreement provides meaningful relief to the affected creators. Over 91 percent of the 482,000 books involved in the claim process have already been registered by their respective authors or publishers for payment.

While the judge ruled that training AI on copyrighted material is not inherently illegal, the legal challenge successfully centered on Anthropic’s acquisition of pirated book copies from unauthorized websites. This outcome represents a notable milestone for copyright litigation in the field of artificial intelligence development.

Legal counsel for the plaintiffs described the outcome as the largest known copyright recovery to date. The distribution process to the affected class members is expected to begin shortly. This case serves as a benchmark for how similar disputes regarding intellectual property and large language models might be resolved in the future.