Human authors pay 1.5 billion dollars to settle a lawsuit for copyright: NPR

In one of the largest copyright settlements involved in obstetric artificial intelligence, AI, a leading company in the field of artificial intelligence, agreed to pay $ 1.5 billion to settle the laws to violate copyright filed by a group of authors.

If the court approves the settlement, the Antarbur will compensate for the authors of about 3000 dollars for each of the 500,000 books covered by the settlement.

The settlement, which will be seen by the Upper Governor in the American province in San Francisco in the approval of next week, in case it included the first fundamental decision on how to apply fair use to obstetric artificial intelligence systems. It also indicates that there is a turning point in the ongoing legal battles between creative industries and artificial intelligence companies accused of using artworks illegally to train large language models that support artificial intelligence systems widely used.

The doctrine of just use allows the use of copyright -protected business by third parties without the approval of the copyright holder in some circumstances, such as clarifying a point in a news article. Artificial intelligence companies are trying to clarify the issue to use the copyright -protected business to train the modular obstetric intelligence models commonly use. But the authors and other prosecutors in the creative industry were retreating.

“This historical settlement will be the largest recovery of copyright and reported in history,” on the pretext that it will “provide meaningful compensation” to the authors and “a precedent for artificial intelligence companies that pay for their use of pirated web sites.”

“This settlement indicates the beginning of the necessary development towards a market -based legitimate licensing system for training data,” said Cecilia Ziniti, a lawyer for the technology industry and the author of the former ninth department that does not participate in this specific issue. “It is not the end of artificial intelligence, but the beginning of a more mature and sustainable environmental system where creators are compensated, like how to adapt the music industry to digital distribution.”

A case with the provisions of the division

Authors Andrea Partz, Charles Gray, Kirk Wallace Johnson presented complaint In contrast to Anthropor due to the violation of copyright in 2024, the collective lawsuit claimed Amnesty International Artificial Intelligence the contents of millions of books protected by copyright to train large language models behind Chatbot, Claude, including at least two works by each prosecutor. The company also bought some printed copies books and wiped it lightly before eating it in its model. The company admitted to doing this amount, which is the fact that the prosecutors raise their complaints. “The anthropoor admitted using a pile to train Claude,” says Al -Shakawi. (Lint is a large and open source database created for training in the large language model.)

“Instead of obtaining permission and paying a fair price for the creations he exploits, anthropologist,” authors ” complaint Countries.

In June to ruleJudge Alsup agreed to the anthropologist, saying that the company’s use of books by the plaintiffs to train their artificial intelligence model was acceptable.

“The use of training was fairly used,” he wrote. “The use of books is discussed to train Claude and its spoilers were very transformative.”

However, the judge has eliminated that the use of Antarbur for millions Pirate Books were not to build their models – books that are copied web sites such as the Libgen Book Library and the Pilimi Library without obtaining the approval of the authors or granting them compensation -. This part of the case was ordered to trial. The judge wrote in the conclusion of his ruling: “We will have a trial on pirated copies used to establish the Central Library of Anthropor and the resulting, actual or legal damages (including for will).” Last week, the parties announced that they had reached a settlement.

The law of copyright in the United States States This deliberate violation of copyright can lead to legal damage of up to $ 150,000 for all the outgoing work. The judge’s order confirms that the anthropoor pore more than 7 million copies of books. So the damage caused by an experiment, if it has passed, can be enormous.

However, Ziniti said that regardless of the settlement, the judge’s ruling actually means that at least in northern California, artificial intelligence companies now have the legal right to train their large language models on copyright -protected work – as long as they get copies of these works legally.

In NPR data, both sides appear satisfied with the result of the case.

“Today’s settlement, if approved, the remaining old claims will be resolved to the plaintiffs,” said Seradhar, Deputy General Adviser to Anthropot. “We are still committed to developing safe systems of artificial intelligence that helps people and organizations to extend their capabilities, enhance scientific discovery, and solve complex problems.”

“This historical settlement is the first of its kind in the era of artificial intelligence,” said Justin Nelson, a lawyer for the authors. “This will provide a meaningful compensation for each category work and put a precedent that requires artificial intelligence companies to pay the owners of copyright. This settlement sends a strong message to AI and creators alike that taking of the copyright works is wrong.”

The creative community responds

The settlement also met with the consent of the creative community.

“This historical settlement is a vital step in admitting that artificial intelligence companies cannot simply steal the creative work of authors to build artificial intelligence just because they need books to develop large high -quality language models,” he said. Authors Syndicate CEO Mary Rasinberger. “We expect the settlement to lead to more licensing that gives authors all compensation and control of the use of their work by artificial intelligence companies, and the situation in the working free market should be.”

“Although the amount of the settlement is very important and represents a clear victory for publishers and authors in the chapter, it also proves what we were saying all the time – artificial intelligence companies can bear compensation for copyright owners for their work without undermining their ability to continue innovation and competition.” Coalition of copyright.

Man is in a good position to deal with great compensation. On Tuesday, the company Declare The completion of a new financing round of $ 13 billion, which amounts to a total value of $ 183 billion.

Meanwhile, the literary world and other parts of the creative sector continue to fight artificial intelligence companies. There have been a large number of lawsuits for violating copyright of artificial intelligence from the artificial intelligence launched by prominent authors, including TA-Nehisi Coates and Comedy Sarah Silverman in recent years. In June, American boycott judge Vince Chapria Grant a dead request To obtain a brief judgment in the Coates and Silverman case against the technology company, which has made an actual limit to this lawsuit. Other continuous cases.

In the latest in a series of legal procedures that involve major entertainment companies, on Friday, Warner Bruce Discovery A. suit At the California Federal Court against Midjourney’s birthplace for the violation of copyright. NPR has arrived at Midjourney to comment.