Anthropic agrees to compensate authors with 1.5 billion dollars to resolve the lawsuit accusing them of using unauthorized chatbot training data
In a groundbreaking development, artificial intelligence (AI) company Anthropic has agreed to pay a staggering $1.5 billion to settle a class-action lawsuit by book authors. The lawsuit, filed by a trio of authors - Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson - who represent a broader group of writers and publishers, accused Anthropic of illegally using millions of copyrighted books to train its AI chatbots.
Anthropic, a privately held company founded by ex-OpenAI leaders in 2021, had been downloading books from pirate websites such as Library Genesis (LibGen) and The Pirate Library Mirror to train its chatbot Claude. However, a federal judge found that while training AI chatbots on copyrighted books isn't illegal, Anthropic wrongfully acquired these books through pirate sites.
The settlement, said to be the largest copyright recovery ever, could mark a turning point in legal battles between AI companies and authors over copyright infringement. As part of the settlement, Anthropic has agreed to destroy the original book files it downloaded.
The Authors Guild, an organisation representing the interests of authors, had expected damages of at least $750 per work and potentially much higher if Anthropic was found to have willfully infringed copyrights at trial. However, the settlement's higher award of approximately $3,000 per work likely reflects a smaller pool of affected books, after taking out duplicates and those without copyright.
Anthropic will pay authors about $3,000 for each of an estimated 500,000 books covered by the settlement. This comes as Anthropic, which put its value at $183 billion after raising another $13 billion in investments, expects to make $5 billion in sales this year, but has never reported making a profit.
The Danish Rights Alliance stated that the settlement would be of little help to European writers and publishers whose works aren't registered with the U.S. Copyright Office. Comparing the AI model to "any reader aspiring to be a writer," the judge found that Anthropic's use of pirate sites raised internal concerns among employees.
The debut thriller novel The Lost Night by Andrea Bartz, a lead plaintiff in the case, was found in the Books3 dataset, a trove of digital books used by AI companies to train their models. Anthropic later shifted its approach and hired Tom Turvey, the former Google executive in charge of Google Books, to help with buying books in bulk and scanning them for its AI model.
This settlement serves as a reminder to AI companies of the importance of respecting copyright laws and obtaining proper permissions for using copyrighted materials. It also underscores the potential financial consequences of failing to do so.
