One reason I’m so anti-generative AI is because many models rely on a data set containing written or visual work used without the original creator’s permission. As a writer myself, it’s a small comfort then that the copyright suit brought against Anthropic has been settled in favour of the author plaintiffs.
Anthropic leveraged a library of millions of pirated books to train its Claude chatbot and, as a result, the AI company is now set to pay out a copyright settlement of $1.5 billion (via Euro News). Some of the affected authors argued this pay out was too small but, as of July 20, a US judge has now approved the deal, closing the book on one of the largest copyright class action lawsuits in US history.
Anthropic agreed to pay the $1.5 billion settlement back in September 2025, but a US judge only approved the deal this week. The settlement covers an approximate total of 500,000 works, with affected authors and publishers to be paid $3,000 for each pirated work. Anthropic says that 91% of those eligible have now filed a claim.
Flicking back to the beginning of this saga, writers Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson originally brought their lawsuit against Anthropic in 2024. The plaintiffs alleged that the AI company had used pirated books to train Claude.