Anthropic AI to settle copyright lawsuit for $1.5B

The largest copyright class action settlement in history received final approval Monday, and artificial intelligence company Anthropic will be the one paying it out. The agreement requires Anthropic to shell out $1.5 billion to the authors and publishers whose work it downloaded from pirate sites. Anthropic used hundreds of thousands of pirated books to train its large-language models. Those AI models use advanced statistics to find patterns in stacks of texts and can become capable of generating content while mimicking human speech or writing.

Each member of the class action lawsuit could receive between $200 and $150,000 per pirated work, according to the settlement. The estimated per-work payment for each piece of copyrighted work will be around $3,000, according to the agreement. Anthropic will also be required to destroy the original files of the pirated works. U.S. District Judge Araceli Martínez-Olguín said in her judgment that the settlement was fair, reasonable, and adequate.

The Association of American Publishers applauded the decision, calling it a victory for the protection of intellectual and creative properties. WORLD reached out to Anthropic for comment but did not immediately hear back.

How did the case get here? A group of authors filed a copyright infringement lawsuit against the company in 2024. A year later, Judge William Alsup ruled that while AI could be trained on copyrighted books, Anthropic had wrongfully and illegally accessed the books. The case was set to go to trial, but the company instead agreed to settle the case in September 2025.