Court gives final approval to Anthropic's $1.5 billion settlement over books pirated to train Claude

What happened: On 20 July 2026, US District Judge Araceli Martínez-Olguín gave final approval to Anthropic's $1.5 billion settlement in Bartz v. Anthropic, the class action brought by authors Andrea Bartz, Charles Graeber, Kirk Wallace Johnson and hundreds of thousands of other writers and publishers. The settlement — the largest copyright class action settlement in US history — pays roughly $3,000 for each of approximately 500,000 works that Anthropic copied from pirate libraries (Library Genesis and Pirate Library Mirror) while developing Claude. Anthropic is required to destroy all the pirated material. More than 91% of covered authors and publishers claimed their share, per Anthropic's statement. Judge William Alsup, who issued preliminary approval before retiring, had split the earlier fair-use ruling: training Claude on legitimately obtained books was 'exceedingly transformative' fair use, but downloading some seven million books from shadow libraries to build a permanent internal repository was not.

Why it matters to creators: This is the first and so far only case in which an AI lab has paid authors directly for books used in model development — a per-work price point (~$3,000) that other litigants and licensors are already citing. It also drew the sharpest legal line to date: the training act may be fair use, but pirating the corpus is not.

Not yet known: Whether any of the still-pending suits against Google, Meta, OpenAI, Midjourney or Perplexity produce comparable payouts, or whether they settle before reaching the 'pirated library' question.