Music Publishers and Anthropic Fight Over AI and Song Lyrics in Court
A California judge is weighing whether an AI company can legally use millions of song lyrics to train its chatbot without paying copyright holders.
A big legal battle is playing out in San Jose, California. On October 21, an AI company called Anthropic asked a judge to protect its right to use millions of song lyrics to train its chatbot, Claude. Major music companies, including Universal Music Group, say Anthropic used their lyrics without permission and broke copyright law. The judge has not made a ruling yet, but the case could affect many other AI lawsuits across the country.
Anthropic's lawyer argued that using song lyrics to train Claude is 'fair use.' Fair use is a rule in US copyright law that lets people use copyrighted material in certain ways without asking permission. Anthropic says its AI training turns those lyrics into something new and different, which is exactly what fair use is meant to protect. The music publishers strongly disagree with this argument.
The publishers' lawyer said Claude received nearly 120,000 requests to create artificial, or 'synthetic,' song lyrics over just six months. He warned that AI tools could flood the music world with machine-made songs, hurting real human songwriters. He said this kind of competition is harmful and is not covered by fair use. The publishers said Anthropic is using their lyrics to create a product that competes against them.
Anthropic fired back by saying there is no proof that Claude has actually hurt the market for song lyrics. Anthropic's lawyer pointed out that Universal Music Group itself has said AI is 'overwhelmingly a net positive' for the music industry. She also said AI has not lowered the demand for real lyrics. Without clear evidence of harm, Anthropic argues there is no reason to hold the company responsible.
Judge Eumi Lee asked tough questions to both sides during the hearing. She seemed especially curious about whether Claude's ability to write lyrics really threatens the publishers' business. Her questions showed she was thinking carefully about both arguments. She has not said when she will make her final decision.
This case is part of a much larger debate happening across the country. Authors, news outlets, and artists have all sued AI companies like Anthropic, OpenAI, and Meta, saying those companies used their work without permission to build AI systems. Two other California judges have already looked at similar cases, and they did not agree with each other. One judge ruled that AI training is fair use, while another warned that it often would not be.
The music publishers who filed this lawsuit — UMG, Concord, and ABKCO — first took Anthropic to court in 2023. They say Claude was trained using lyrics from songs by famous artists like Beyoncé, the Rolling Stones, and the Beach Boys. Anthropic says it used those lyrics only to help Claude understand human language and to support progress in science, business, and education. Whatever Judge Lee decides could shape the rules for AI companies for years to come.
Large language models like Claude "could overwhelm the market with AI-generated content very easily."
Comprehension quiz preview
1. Where did the court hearing in this case take place?
2. How many requests did the publishers' lawyer say Claude received to generate synthetic lyrics over six months?
3. Which music publishers sued Anthropic in 2023?