Meta has scored a fair use win in the AI training lawsuit levied against it by multiple authors. Photo Credit: Dima Solomin
Another day, another fair use ruling – this time in favor of Meta, which has scored a summary judgement win in the copyright lawsuit filed against it by authors including Sarah Silverman and Richard Kadrey.
Judge Vince Chhabria handed down that ruling yesterday, after previously dismissing a portion of the long-running action. As we’ve reported – and as many are aware – said action centers on Meta’s allegedly unlawful training of Llama on protected books.
Of course, this is one of several complaints – filed by different authors, music companies and professionals, and others yet – centering on gen AI training. In general, developers are adamant that ingesting all manner of protected works without authorization constitutes fair use.
Judge Alsup just recently agreed with that assessment in copyright litigation levied against Anthropic. This training process (besides breaking down and scanning physical books into a central digital database) is wholly above board, the court found in more words.
That said, while Judge Chhabria’s ruling in the authors v. Meta suit represents another legal setback for rightsholders, the precedent at hand isn’t nearly as sweeping. In fact, the judge determined that mass-training, though transformative, is “likely” illegal “in most cases.”
“There is certainly no rule that when your use of a protected work is ‘transformative,’ this automatically inoculates you from a claim of copyright infringement,” Judge Chhabria wrote, also directly addressing the decision in the mentioned Anthropic suit. “And here, copying the protected works, however transformative, involves the creation of a product with the ability to severely harm the market for the works being copied, and thus severely undermine the incentive for human beings to create.
“Under the fair use doctrine, harm to the market for the copyrighted work is more important than the purpose for which the copies are made,” he continued.
Similarly, the court refuted the “ridiculous” idea that rulings in favor of rightsholders will stunt the growth trajectory of LLMs. “These products are expected to generate billions, even trillions, of dollars for the companies that are developing them,” he stated. “If using copyrighted works to train the models is as necessary as the companies say, they will figure out a way to compensate copyright holders for it.”
Why, then, did Judge Chhabria grant Meta’s motion for summary judgement? In greater detail, the court criticized the case’s two chief arguments – that Llama can produce snippets of the involved books and has harmed the plaintiffs’ ability to license their works for gen AI – as “clear losers.”
“Llama is not capable of generating enough text from the plaintiffs’ books to matter, and the plaintiffs are not entitled to the market for licensing their works as AI training data. As for the potentially winning argument—that Meta has copied their works to create a product that will likely flood the market with similar works, causing market dilution—the plaintiffs barely give this issue lip service,” the judge explained.
(Needless to say, the legal system isn’t exactly on the same page when it comes to heretofore unencountered AI questions. Despite its broad view of fair use for actual LLM training, the court struck a different tone concerning possibly infringing outputs in the Anthropic suit. “Here, if the outputs seen by users had been infringing, Authors would have a different case,” Judge Alsup penned. “And, if the outputs were ever to become infringing, Authors could bring such a case.”)
Back to the Meta suit, Judge Chhabria described the ruling’s consequences as “limited” in the bigger picture.
“And, as should now be clear,” he proceeded, “this ruling does not stand for the proposition that Meta’s use of copyrighted materials to train its language models is lawful. It stands only for the proposition that these plaintiffs made the wrong arguments and failed to develop a record in support of the right one.”
Moving forward, it’ll be worth closely monitoring the determination’s impact – particularly when it comes to different plaintiffs’ “meaningful evidence on market dilution” stemming from unauthorized training and then outputs – on future litigation.
As things stand (and similarly to the Anthropic case), the claim that Meta illegally distributed the authors’ “works during the torrenting process” is still alive; a case management conference is set for July 11th.
