Artificial intelligence company Anthropic has agreed to pay at least $1.5 billion (€1.3 billion) to settle a class action case brought by authors who accused the firm of exploiting their work without permission.
The proposed deal, filed in court on Friday, still requires judicial approval before payments are released.
The lawsuit was launched last year by writers Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson.
They alleged that Anthropic used pirated copies of books to train its chatbot Claude, which competes directly with OpenAI’s ChatGPT.
The case quickly grew into one of the most watched disputes in the clash between copyright and artificial intelligence.
According to court records, the settlement covers around 500,000 books, with roughly $3,000 (€2,500) set aside for each title.
This makes it the largest recovery to date in disputes over copyrighted material used for AI training.
The Authors Guild and other advocacy groups welcomed the outcome, saying it marks a turning point for creative professionals whose work has been used in datasets without payment or consent.
Analysts note that the case could shape the direction of other high-profile lawsuits already filed against OpenAI and Meta.
Publishers and writers continue to argue that the use of copyrighted works in AI systems undermines intellectual property rights. Technology companies, however, maintain that such training practices should be protected under fair use principles.
If approved, the Anthropic settlement will set a benchmark for how disputes between AI developers and content creators are resolved, likely pushing other firms to negotiate rather than face billion-dollar penalties.