A federal judge sat with Meta on Wednesday in a lawsuit brought against the company of 13 book authors, including Sarah Silverman, who claimed that the company illegally trained its AI models on their copyrighted works.
Federal Judge Vince Chhabria issued Summary judgment – meaning that the judge could decide on the case without sending it to a jury – in favor of Meta, finding that the AI Company’s training models in this case fell under the “fair use” doctrine of copyright law and was thus legal.
The decision comes only a few days after A federal judge was sitting with anthropic in a similar complaint. Together, these cases form to be a victory for the TE industry industry, which has spent years in legal battles with media companies arguing that training of AI models on copyrighted works are fair use.
However, these decisions are not the sweep wins some companies – both judges have realized that their cases were limited to amplitude.
Judge Chhabria explained that this decision does not mean that all AI model trainings on copyrighted works are legal, but rather that the plaintiffs in this case have “made the wrong arguments” and have not been able to develop sufficient evidence to support the right one.
“This decision is not aimed at the proposal that the use of meta -copyrighted materials to train their language models is legal,” Judge Chhabria said in his decision. Later, he said, “In cases involving uses like Meta, it seems that the plaintiffs will often win, at least where those cases have better developed records on the market effects of the defendant’s use.”
Judge Chhabria ruled that the use of copyrighted works by Meta in this case was transformative – that means that the company’s AI models not only reproduced the authors’ books.
Further, the plaintiffs failed to convince the judge that the copying of the books has damaged the market for those authors, which is a key factor to determine whether copyright law was violated.
“The plaintiffs presented no significant evidence of market dilution,” said Judge Chhabria.
Both Antropic and Meta wins involve book training models, but there are several other active processes against technology companies for training AI models on other copyrighted works. For example, The New York Times asks Openai and Microsoft for training AI Models on news articles, while Disney and Universal ask Midjourney for training Models.
Judge Chhabria noted in his decision that fair use defenses depend much on the details of a case, and some industries may have stronger fair arguments than others.
“It seems that markets for some types of works (like news) could be even more vulnerable to indirect competition from AI results,” said Chhabria.
