Humanity, an AI company, agreed to a $1.5 billion settlement on Friday. It aims to resolve a sweeping class action lawsuit filed by the author who allegedly used pirated copies of the book to train the chatbot Claude.
According to the proposed agreement, subject to judicial approval, the settlement offers approximately $3,000 for each of the authors for an estimated 500,000 books covered. If approved, it will be ” The largest publicly reported copyright reconstruction in history. ”
The author guild, representing thousands of writers, welcomed the results. On Friday, CEO Mary Raysenberger said the settlement was “a great outcome for authors, publishers and right-wingers, sending a strong message about the serious consequences when pirated authors' works to train AI, taking away those who can't afford it.”
The lawsuit was launched last year by author Andrea Burtz and authors Charles Graeber and Kirk Wallace Johnson. They later became representative of a wider class of writers and publishers after being accused of downloading millions of pirated books to train AI models.
In June, US District Judge William Alsap ruled that while training AI chatbots with copyrighted books is not illegal in itself, humanity misrepresented more than 7 million digitized works from copyrighted websites, including Books3, Library Genesis and The Pirate Library Mirror.
If the company goes to trial and loses in December, analysts said the financial blow could be devastating. “We're looking forward to seeing you in the future,” said William Long, a legal analyst at Wolters Kluwer.
This transaction comes amidst increased scrutiny among AI companies. Last month, X Corp and X.ai submitted antitrust laws against Apple and Openai, denounced the exclusive practices on smartphones and generator AI chatbots. Additionally, in Texas, Attorney General Ken Paxton recently began investigating meta and characters. As to whether their chatbots misinterpreted their children with deceptive claims that they provide therapeutic support and raise concerns about privacy and data exploitation.
