Judge Dismisses Parts of Sarah Silverman's AI Copyright Case Against Meta

Core argument that Meta used copyrighted books to train its LLaMA language models remains unchallenged.

Parts of Sarah Silverman's AI copyright case against Meta have been dismissed by a federal judge, but the core argument remains unchallenged.
The plaintiffs had accused Meta of improperly using their copyrighted books to train its LLaMA language models.
The claims dismissed were broad arguments about the scope of Meta’s alleged liability, including that every LLaMA output was an infringing derivative work.
The judge ruled that the LLaMA models themselves cannot be seen as a recasting or adaptation of any of the plaintiffs’ books.
The authors will have 21 days to file an amended complaint.