Humanity agreed An estimated $3,000 per work to pay at least $1.5 billion to resolve a lawsuit filed by a group of authors of books claiming copyright infringement. In a court motion Friday, the plaintiffs stressed that the terms of the settlement were “significant victory” and that going to trial is a “major” risk.
This is the first class of action settlements centered around American AI and copyright, and the results could shape how regulators and the creative industry approach legal debates about generated AI and intellectual property. According to the settlement agreement, class action lawsuits apply to around 500,000 tasks, but the number could rise once the list of pirated materials is finalized. For each additional work, the artificial intelligence company pays an additional $3,000. The plaintiff plans to provide the court with a final list of works by October.
“This groundbreaking settlement far outweighs other known copyright restorations. This is the first time in the AI era. It provides meaningful compensation for each class of work and sets precedents that require AI companies to pay the copyright owners. LLP.
Humanity does not recognize fraud or liability. “Today's settlement will resolve the remaining legacy claims of the plaintiffs, if approved. We are still committed to developing secure AI systems that will help people and organizations expand their capabilities, advance scientific discoveries, and solve complex problems.”
The lawsuit originally filed in the U.S. District Court for the Northern District of California in 2024 was part of a larger ongoing wave of copyright lawsuits brought against high-tech companies over the data they used to train artificial intelligence programs. Authors Andrea Bartz, Kirk Wallace Johnson, and Charles Graeber argued that humanity trained large-scale language models in their work without permission, violating copyright laws.
In June this year, Senior District Judge William Alsup ruled that human AI training is protected by the doctrine of “fair use.” It was a victory for tech companies, but there was a big warning. When gathering materials to train AI tools, humanity relied on a corpus of pirated books from the so-called “shadow libraries,” including the infamous site Libgen. (Humanity claims that he did not actually train the product with pirated works and instead chooses to purchase a copy of the book.)
“Humanity has downloaded more than seven million copies of famous books, paid nothing, and keeps these pirated copies in its library.
