American actress and comedian Sarah Silverman has faced a setback in her legal battle against Facebook parent company Meta regarding artificial intelligence (AI). Silverman filed a lawsuit alleging that Meta used her copyrighted books without authorisation for training its generative AI model.
However, US District Judge Vince Chhabria dismissed a significant portion of Silverman’s lawsuit, labelling some of the claims as, “Nonsensical”. “There is no way to understand the LLaMA models themselves as a recasting or adaptation of any of the plaintiffs’ books,” Chhabria wrote in his ruling, via The Hollywood Reporter.
The judge also dismissed arguments that any outputs from Meta’s AI tool would amount to copyright infringement, citing the lack of evidence that these results could be interpreted as reinterpreting, transforming or adapting the books in question.
Silverman’s lawsuit argued that the use of her intellectual property led to the creation of infringing derivatives, as these companies stood to gain from every search conducted by third-party users. The lawsuit also alleged that the AI companies violated competition laws.
A lack of proof
“To prevail on a theory that LLaMA’s outputs constitute derivative infringement, the plaintiffs would indeed need to allege and ultimately prove that the outputs ‘incorporate in some form a portion of the plaintiffs’ books,” wrote Chhabria.
This lawsuit is not the first of its kind to assert that using intellectual property, such as books, art, and images, to train AI bots constitutes copyright infringement. Silverman became part of a class action making similar claims against Meta in July. However, US District Judge William Orrick also dismissed numerous arguments in that lawsuit.
He characterised the arguments as, “Defective in numerous respects” due to the plaintiffs’ inability to provide evidence of material produced by AI. That legal action also targeted StabilityAI, DeviantArt, and Midjourney.
While facing copyright infringement claims, Meta and other AI companies maintain that their use of copyrighted material is protected under fair use principles. In light of the two recent court decisions, plaintiffs alleging copyright infringement will likely face the challenge of presenting evidence where AI-generated content directly mirrors the original copyrighted material.