A federal judge in California temporarily dismissed a copyright lawsuit against Meta Platforms (META) filed by comedian Sarah Silverman and other authors concerning its Llama artificial intelligence program.
During a hearing, U.S. District Judge Vince Chhabria supported Meta’s motion to dismiss the writers’ claims that Llama-generated text infringes on their rights. Judge Chhabria also indicated that he would allow the authors to amend most of their statements.
The authors’ primary argument, alleging that Meta infringed on their rights by utilizing their books to train Llama, remains unchallenged.
Addressing the scholars’ legal representatives, Judge Chhabria stated, “I grasp your fundamental argument. However, I find the rest of your legal grounds unclear.”
This lawsuit is just one among several legal actions where copyright holders are suing technology companies over the use of AI algorithms. The central issue revolves around the legality of employing copyrighted material to train AI models without authorization.
In a previous lawsuit filed in July, the authors sued Meta and Microsoft-backed OpenAI, claiming that the companies breached their copyrights by utilizing their books to train AI language models, including the resulting text.
Judge Chhabria questioned the assertion and pondered whether the content generated by Llama closely resembled or replicated the authors’ works.
He remarked, “When I interact with Llama, I am not seeking a passage from Sarah Silverman’s book.”
The authors argue that Llama itself infringes on intellectual property rights. Judge Chhabria suggested that the argument “would imply that you could argue they are similar if you juxtaposed Llama’s output with Sarah Silverman’s book.”
He added, “That concept is mind-boggling.”
Unless the authors can demonstrate a significant similarity between Llama’s output and their works, Judge Chhabria warned that he might dismiss most of the claims with the opportunity to amend them.
Separately, OpenAI requested that the court revise the authors’ lawsuit against them from August, with a hearing scheduled for the upcoming month.
In a related development, a federal judge in California recently rejected similar claims brought by visual artists against Stability AI, Midjourney, and DeviantArt. The judge concluded that the artists’ creations were likely not substantially akin to the AI-generated content to constitute infringement.
