FOLLOW US

A California federal judge indicated on Thursday that she is leaning toward dismissing parts of a copyright lawsuit filed by music publishers against Anthropic PBC, the AI company accused of training its chatbot, Claude, using copyrighted song lyrics. The case, which has garnered attention for its novel legal implications, includes allegations of contributory infringement, vicarious infringement, and the unlawful removal or alteration of copyright management information.

U.S. District Judge Eumi K. Lee, presiding over the case, said during a Zoom hearing that some of the claims made by the music publishers were "so general" and might require further clarification. She tentatively ruled in favor of allowing the plaintiffs to amend their complaint. Importantly, Anthropic's motion does not address the primary claim of direct copyright infringement, which remains intact.

Concerns Over Generalized Allegations

During the hearing, Judge Lee noted her concern about the broad nature of some of the allegations but expressed confidence that the issues could be addressed through amendment.

"The complaint seems like it is amendable," Judge Lee remarked, highlighting that specificity is critical, especially at this stage of litigation.

Nicholas Hailey of Oppenheim & Zebrak LLP, representing the music publishers, argued that the novel nature of the case should be taken into account when evaluating the sufficiency of the claims.

"The novelty of this case weighs strongly against dismissal at the motion to dismiss stage," Hailey said.

However, Judge Lee reiterated that the allegations for contributory and vicarious infringement appeared to lack the necessary specificity. She criticized the use of phrases such as "it could" or "would," which she found overly speculative.

Evidence and Guardrails

Hailey contended that the complaint sufficiently supports the claims for third-party infringement, pointing to evidence that Anthropic’s Claude provided copyrighted song lyrics to the plaintiffs' investigators. He argued that this behavior suggests the chatbot may have done the same for other users.

For example, he said, when prompted to create a song about Buddy Holly’s death, Claude generated lyrics from "American Pie" while claiming the output was original.

Anthropic had also implemented guardrails to prevent copyright infringement, Hailey noted, which he argued demonstrates the company’s awareness of the issue. "These guardrails were largely ineffective, but they show Anthropic knew there was an issue," he said.