On October 21, a U.S. federal court in California heard a copyright dispute between Anthropic and music publishers. Concord, Universal Music Publishing Group, and others accused Claude of using copyrighted song lyrics for training without authorization and potentially directly outputting original lyrics, eroding commercial lyric licensing revenue. Another related lawsuit involves 20,517 musical works.
The plaintiffs are seeking statutory damages of up to $150,000 for each count of willful copyright infringement. Anthropic argues that AI training constitutes fair use and emphasizes that its model has copyright protection mechanisms. At the hearing, the judge focused on whether AI-generated lyrics harmed the potential market for the original works, and no final ruling has been made.
It is worth noting that Anthropic previously reached a $1.5 billion settlement in a book piracy lawsuit. This dispute may further clarify the boundaries of fair use in AI training, liability for obtaining pirated data, and standards for model output infringement, impacting data licensing costs across the entire generative AI industry.
