Piracy has become a hot-button issue in AI litigation ever since Anthropic reached a $1.5 billion settlement for torrenting books — but such claims won’t be part of the copyright infringement lawsuit brought by music publishers against the Claude chatbot maker.
Federal court records show that on Wednesday (Oct. 8), Judge Eumi K. Lee denied a motion by Universal Music Publishing Group, Concord Music and ABKCO Music to amend their 2023 lawsuit against Anthropic. The case centers on claims that Anthropic trained its large language model Claude on copyrighted song lyrics and now spits out copycat lyrics when prompted.
The music publishers asked in August to add piracy claims to the lawsuit, saying they’d recently learned that seven million books in Anthropic’s training arsenal were illegally torrented from online pirate libraries. The publishers said their intellectual property is likely part of this booty, including numerous music and lyrics compendiums and the sheet music to hits like Elton John’s “Tiny Dancer” and Ariana Grande’s “7 Rings.”
The publishers’ proposed amended complaint was part of a bigger trend in the dozens of AI copyright cases making their way through the court system. Plaintiffs have been focusing on piracy since a landmark June court opinion held Anthropic liable for storing torrented books, and the trend only intensified after Anthropic settled that case for $1.5 billion in September.
The major record labels, for example, are currently trying to amend their copyright lawsuits against AI music generators Suno and Udio with new stream-ripping claims. Suno and Udio say such allegations are legally bogus.
Anthropic opposed the amendment in the music publishers’ case as well, slamming the new claims as an improper attempt to “fundamentally transform this case at the eleventh hour.” The AI company said the publishers could have discovered their torrenting activity long ago, and that adding piracy claims now would unfairly delay the already two-year-old case.
Related
Judge Lee ultimately sided with Anthropic. According to Bloomberg Law, the judge stated during a court hearing in San Jose, Calif., on Wednesday that the amendment is inappropriate because the publishers failed to properly investigate the piracy theory earlier.
The publishers’ lawyers declined to comment on the ruling on Thursday (Oct. 9). Anthropic’s reps did not immediately return a request for comment.
While Wednesday’s ruling is a loss for the publishers, they won a victory just a few days earlier when Judge Lee rejected Anthropic’s bid to trim down the original copyright infringement claims on Monday (Oct. 6).
Anthropic insists that training Claude on copyrighted material is fair use — a foundational tenet of copyright law that allows protected works to be recycled for “transformative” purposes, like news reporting or parody.
Whether or not AI training constitutes fair use is a hotly contested legal question that is still being hammered out by the courts. This issue is the subject of ongoing evidence discovery in the publishers’ case against Anthropic, which is scheduled to run through March.
