Anthropic’s $1.5B book settlement: a win for authors or a tax on training?
a judge approved anthropic's $1.5b settlement for training on pirated books, paying authors $3,000 per work. this establishes a massive precedent for ai copyright liability.
judge araceli martínez-olguín approved a $1.5bn settlement between anthropic and a class of authors, resolving allegations that the company illegally trained its ai models on copyrighted books. under the terms, authors receive approximately $3,000 per infringed book, a record-setting figure that anthropic's general counsel says over 91% of claimants have accepted. while this closes a major legal chapter for the company, anthropic still faces separate lawsuits from authors arguing the damages are inadequate. this ruling signals the high cost of unconsented data use and effectively sets a market rate for book licensing in ai training datasets.