Meta Wins In Illegal AI Training Lawsuit With Copyright Rights Book

JAKARTA In 2023, Meta was sued by dozens of writers on charges of illegally training the Artificial Intelligence (AI) system. In that case, the judge finally decided to support Meta.

Judge Vince Chhabria granted the request for a summary decision submitted by Meta. The judge also stated that the use of authors' copyrighted works to train Meta's AI model is a fair use or reasonable use.

"(Meta) has the right to accept a summary verdict on its reasonable use of claims that copying these plaintiffs' books to be used as LLM training data is a violation," Chhabria said, quoted by The Verge on Thursday, June 26.

Judge Chhabria explained that they sided with Meta because the plaintiff failed to present strong arguments. The federal judge also stressed that the authors failed to build enough records to support their claims.

"That only applies to the proposition that this plaintiff made wrong arguments and failed to develop records to support correct arguments," added judge Chhabria in his decree issued last Wednesday, June 25.

The two main arguments put forward by a dozen authors in the Meta case are considered 'severe losses' by Chhabria. The first argument alludes to Meta's Llama AI ability in reproducing some of the text from their books

=-= read

Meanwhile, the second argument notes that Meta uses their work without permission, thus undermining their potential to license their work as training data.

The judge stated that Llama could not produce enough text from the plaintiff's book to be a problem. He also emphasized that the plaintiff has no right to the market to license their work as AI training data.

Chhabria also touched on Anthropic's decision, in which Judge William Alsup ignored concerns about the potential dangers of a generative AI in the markets of his trained works. This decision further strengthens Meta's position in using copyrighted data for training.