Mega score a The company made a major victory in Wednesday’s copyright lawsuit when a federal judge ruled that the company trained its AI tools on 13 unauthorized books of authors.
“The court has no choice but to give a summary judgment on Meta on the plaintiff’s claim that the company violated copyright law by training its models with books,” U.S. District Judge Vince Chhabria wrote in the summary judgment. He concluded that the plaintiff did not provide enough evidence that Meta’s use of the book was harmful.
In 2023, a group of high-profile writers, including comedian Sarah Silverman, sued Meta for alleging that the tech behemoth had violated their copyrights and by performing large language models on their work. Kadrey v. Meta It is one of the first cases of similar situations; now, there are dozens of similar AI copyright lawsuits through U.S. courts.
Chhabria had previously stressed that he planned to carefully examine whether the plaintiffs had enough evidence that Meta’s use of the job would financially harm them. “In any case, the key issue in which the defendant reproduces someone’s original work without permission is whether allowing people to participate in such behavior will greatly reduce the market of the original market,” he wrote in his judgment on Wednesday.
This is the second major ruling in the AI copyright community this week; on Monday, Judge William Alsup ruled that it is legal to anthropomorphize using copyright-protected materials to train your own AI tools. Chhabria mentioned Alsup’s summary judgment in his decision.
Chhabria painfully stressed that his ruling was based on specific facts in this case, which allowed other authors to open up lawsuits to make future copyright infringement. “In the grand plan of things, the consequences of this ruling are limited. This is not a class action lawsuit, so the ruling will only affect the rights of these 13 authors, not the works that countless others use to train their models,” he wrote. “And, as it should be clear now, this ruling does not represent the legality of the claim that meta-elements train their language models using copyright-protected materials.”
This is a developing story. Please check for updates.
