Anthropic has won final approval for a $1.5 billion class-action settlement with thousands of authors and publishers over training its Claude AI on copyrighted works, with payouts of about $3,000 per work. Bloomsbury, publisher of Harry Potter, is among the beneficiaries, expected to receive about $19 million after fees, with 14,087 titles covered. The settlement resolves past use of pirated libraries to assemble training data, while affirming that the underlying training of AI on copyrighted text is fair use, and it does not shield Anthropic from future claims. Critics note the deal still leaves broader legal questions unresolved and that damages per title are modest given the volume of works involved. The court also required the destruction of pirated files and clarified that future output-based claims remain open, marking the case as the largest known copyright recovery in history but not a universal resolution of AI copyright disputes.