Meta and OpenAI Use of Copyrighted Books for Training AI Was Fair Use: Federal Judge

A federal court yielded an important blow this week for authors who sue technical giants about AI training. The judge ruled that the use of meta of copyrighted books to train his artificial intelligence models formed reasonable use under copyright legislation.

Judge Vince CHABRIA in the US in San Francisco chose the side of Meta platforms on Wednesday in a case that was brought by 13 authors, including comedian Sarah Silverman and Pulitzer Prize Winners Junot Díaz and Andrew Sean Greer.

The 13 authors who suggest Meta did not provide sufficient evidence that the AI ​​of the company would dilute the market for their work, Judge Chababria said in the pronunciation.

Their argument, he said, “this issue hardly gives any lip service” and there was no facts that are needed to prove damage under the US copyright legislation.

But the judge made it clear that the ruling is far from a general approval of the controversial training practices of AI companies.

“This statement does not stand for the statement that the use of meta of copyrighted materials to train its language models is legal,” said Chabria. “It only stands for the statement that these claimants have made the wrong arguments and have not developed a record to support the right one.”

Kunal Anand, CEO of Ai Chatbot Service Aibaat, said Decrypt That he hopes it is a sign that courts will find a way to “balance technological progress with the rights of the maker.”

“Although the decision was preferred to Meta, it reminds us that ethical AI development requires clear licensing frames,” he added.

The authors complained Meta and OpenAi in 2023 and claimed that the companies have abused illegal versions of their books to train their Lama AI and Chatgpt systems without permission or compensation.

In January, the court applications revealed that Meta -CEO Mark Zuckerberg is personally approved with the help of the illegal data set, despite warnings from his AI team that it was obtained illegally. Encouraged internal messages in the archives showing engineers of Meta hesitates, in which an employee admits: “Torrenting of a business laptop does not feel good.”

But the company continued.

Judge Chabria acknowledged the potential for AI to “flood the market with endless quantities of images, songs, articles, books and more” with “a small fraction of the time and creativity that would otherwise be needed.”

He noted in the statement that “this could drastically undermine the market for those works, and therefore dramatically undermine the stimulus for people to create things in the old -fashioned way.”

Chahabria expressed sympathy for the concerns of authors, but it was not enough to make a good legal argument. “Courts cannot decide things based on general insights,” he said.

The prevailing preference for Meta only affects these 13 specific authors, because it was not certified as a Class Action.

This week, the decision marks the second big win for AI companies, after a similar prevailing preference for anthropic on Monday.

In that case, judge William ALSUM also found ai -training reasonable use, but criticized anthropic for building a permanent library with illegal books.

Experts say that the solution for disputes about AI training and copyrighted content lies in proactive market-based approaches instead of waiting for the clarity of the regulations.

“By the time that policy makers overtake the latest AI breakthroughs, those breakthroughs will have demanded a different generation,” said Hitesh Bhardwaj, co-founder at Capx Ai, said Decrypt. “A more sustainable path is to reward people whose work AI assumes: create transparent market places where authors and makers give their own data on fair conditions in license.”

“That approach puts control back in the hands of the people whose content drives our models,” he said.

Published by Stacy Elliott.