Anthourbroid agrees to the authors for at least $ 1.5 billion in the settlement of publishing rights. Amnesty International

Human agreed To pay at least $ 1.5 billion to Law of the lawsuit It was presented by a group of books who claim to violate copyright, estimated at $ 3,000 per work. In a request from the court on Friday, the prosecutors confirmed that the conditions of the settlement are “critical victories” and that going to the trial would be a “enormous” risk.

This is the settlement of the first lawsuit that focuses on artificial intelligence and copyright in the United States, and the result may be how organizers and creative industries deal with legal debate about artificial intelligence and intellectual property. According to the settlement agreement, the collective lawsuit will be applied to about 500,000 works, but this number may rise as soon as the list of pirated materials is completed. For every additional work, the artificial intelligence company will pay an additional $ 3000. Prosecutors are planning to submit a final list of works to court by October.

“This historical settlement greatly exceeds any other recovery known for copyright. It is the first of its kind in the era of artificial intelligence. It will provide meaningful compensation for each separation work and put a precedent that requires artificial intelligence companies to pay copyright. Godfrey LLP.

Man does not recognize any violations or responsibility. “If the settlement of today, if it is approved, it will solve the remaining old claims for the prosecutors. We are still committed to developing safe systems of artificial intelligence that helps people and organizations to expand their capabilities, develop a scientific discovery, and solve complex problems,” said Deputy Attorney General of the Alabaro Sariderhar in a statement.

The lawsuit, which was originally filed in 2024 at the American District Court of the Northern Region in California, was part of a More continuous wave It is a copyright that was presented against technology companies about the data they used to train artificial intelligence programs. Authors Andrea Partz, Kirk Wallace Johnson, and Charles Grayber claimed that the Antarbur trained its great language models for their work without permission, which violates the law of copyright.

In June, the senior boycott judge, William Al Sap Ruling Anthropor’s training was protected from artificial intelligence through the doctrine of “fair use”, which allows unauthorized use of publication and copyrights under certain circumstances. It was a victory for the technology company, but it came with a great warning. While collecting materials to train artificial intelligence tools, Antarbur relied on a group of books pirated from the so -called “shadow libraries”, including A reputable site Libgen, and Alsup decided that the authors should be able to bring the anthropology to the trial in a collective action on limiting their business. (Antarbur confirms that she has already trained her products on pirated business, and instead she chooses to buy copies of books.)

“Anthropor has been downloaded more than seven million pirated copies of books, and did not pay anything, and these pirated copies have kept in her library even after she decided that she would not use it to train artificial intelligence (at all or again). The authors have argued that he should have been pushing anthropiser for the pirated library. This matter agrees.”