The claim and the catalog
Authentic Brands Group is seeking $150,000 for each of the 274 songs, or about $41 million in total. The works named include “Can’t Help Falling in Love,” “All Shook Up” and “A Little Less Conversation.”
The suit follows copyright claims against AI companies from the Recording Industry Association of America and labels including Sony, Universal and Warner Records. But it is among the first major cases brought directly by an artist’s heirs against an AI developer, giving the dispute a more specific test: whether rights holders can challenge the use of an established music catalog in model training.
Lead plaintiffs’ lawyer Richard Busch wrote that the plaintiffs seek to protect exclusive rights from a defendant whose business model, they allege, is built entirely on copyright infringement at an unprecedented scale.
The plaintiffs also allege that Claude has generated infringing outputs “virtually” for every song in any catalog available online since the model launched. That is an allegation in the filing, not a finding by a court.
The unresolved question is control
The complaint says Anthropic introduced safeguards against music-copyright infringement only after it was caught, and that the measures worked poorly. The plaintiffs allege the company knew the protections did not prevent infringement and knowingly continued to violate music copyrights.
The case lands amid broader concern from musicians that models use their music and voices without permission, and that low-quality AI music could displace human work. Some AI bots have already impersonated living artists on streaming services. Last week, Spotify said it had removed about 50 million tracks it considered low-quality AI-generated music from user recommendations over the past year.
I think the case’s most consequential issue may be less the number of songs than the gap between a safeguard and a guarantee. Music-generation platforms analyze large volumes of recordings to learn sound, structure and expression, then let users create music with text or voice prompts. The complaint argues that Anthropic’s protections did not stop copyright violations; what it does not establish is how a court should judge whether a model’s safeguards are sufficient.
That question matters beyond one artist’s catalog. If the claim succeeds, it could set a precedent for AI companies’ use of existing music catalogs. If it does not, the dispute over training rights and the limits of output controls will remain unresolved.
Daily AI news
Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.
Only what matters — every day
Follow on X