Written by Haim Ravia and Dotan Hammer
A California federal judge has dismissed claims that Meta’s Large Language Model (LLaMA) constitutes a derivative work of the materials it was trained on, labeling the claim as “nonsensical.” The lawsuit, asserted by comedian Sarah Silverman and others, challenged Meta’s use of their books for AI training but had most claims dismissed. However, the judge allowed the plaintiffs to amend certain claims and the direct copyright infringement claim over unauthorized copying of books will continue.
The court rejected the plaintiffs’ argument that the models constitute infringing derivative works because they use content from the plaintiffs’ books. The court emphasized that LLaMA models do not reconstitute or adapt the books, which is a prerequisite for classifying a work as derivative. The lawsuit’s claims under the Digital Millennium Copyright Act (DMCA) were also dismissed because there was no indication that LLaMA had distributed the plaintiffs’ books. The court also dismissed the unjust enrichment and negligence claims, the latter dismissed under the economic loss doctrine, which restricts recovery for purely economic losses in tort actions.
A lawsuit against DoNotPay Inc., an AI-based legal service provider which describes itself describes itself as “The World’s First Robot Lawyer”, was also recently dismissed. MillerKing LLC, a law firm, filed a lawsuit against DoNotPay. The lawsuit focused on allegations of false association and false advertising under the Lanham Act and Illinois state law. The lawsuit also accused DoNotPay of unauthorized practice of law. The court dismissed the lawsuit for lack of standing. MillerKing failed to prove any harm or business loss due to DoNotPay’s conduct, a crucial element for establishing standing in federal court.
Click here to read the court order in RICHARD KADREY, et al., v. META PLATFORMS, INC.
Click here to read the court order in MILLERKING LLC v. DONOTPAY INC.
