One of the tactics that plaintiffs are now trying to use in some of the AI copyright litigation is an attempt to pierce the attorney-client privilege in communications related to the AI company’s use of datasets from shadow libraries.
This plaintiffs’ strategy has been deployed in Kadrey v. Meta, Bartz v. Anthropic, and In re OpenAI Copyright Infringement Litigation. In the last case, Magistrate Judge Wang is considering the loss of privilege by OpenAI based on its purportedly shifting explanations for destroying (or not) the Books1 and Books2 datasets.
Earlier this year, Judge Chhabria rejected Kadrey’s attempt to invoke the crime-fraud exception after the Judge reviewed in-camera 50 documents of Meta’s containing attorney-client communications. But Judge Chhabria found no crime or fraud, and rejected Kadrey’s motion.
Well, it was surprising to see Kadrey attempt this strategy again with a different document apparently related to Meta’s torrenting of files. Magistrate Judge Hixson rejected it. But he did so on the narrower ground the communication in question occurred after, and not to facilitate, the alleged crime or fraud and therefore did not provide any basis for loss of privilege.
