We are just beginning to wade into the class action issue in the copyright litigation against AI companies.
Judge Lee just granted Google’s motion to strike the proposed class allegation by the Leovy and Zhang plaintiffs because it was an inappropriate “fail-safe” class, whose definition hinged on anyone who copyright was violated by Google in its training of its AI models.
But Judge Lee also allowed the plaintiffs leave to amend their complaint to allege something the plaintiffs proposed:
“Here, the issue can be resolved by adopting the class definition that Plaintiffs proposed in their stipulations. The new class definition would include: “All persons or entities domiciled in the United States who owned a United States copyright in any work used by Google to train Google’s Generative AI Models during the Class Period.” Redefining the class in this manner is appropriate because class membership can be determined according to objective criteria – that is, ownership of a copyrighted work that was used by Google to train its generative AI models. Membership in Plaintiffs’ new class definition would not require an upfront determination by the Court that each potential class member will prevail on the merits of an infringement claim. See Kamar, 375
F. App’x at 736.”
More of Judge Lee’s opinion is below. In a different case, Anthropic recently filed its opposition to class certification in a brief that is heavily redacted.
