The owners of Elvis Presley‘s music catalog filed a lawsuit on Friday, claiming that Anthropic, the company behind the chatbot Claude, incorporated 274 of the singer’s songs into its training data without obtaining any licenses. Among the titles listed are iconic numbers such as “Can’t Help Falling in Love,” “A Little Less Conversation,” and “Don’t Be Cruel,” as well as “All Shook Up,” “Return to Sender,” and “(You’re The) Devil in Disguise.” The complaint alleges that the AI system now generates text that reproduces lyrical structures and melodies that are substantially similar to the original works.
Legal claims and demanded penalties
Lead counsel Richard Busch framed the case as a defense of exclusive rights, stating that Anthropic’s entire business model rests on what he described as “copyright infringement at a scale never seen before.” The suit seeks statutory damages of $150,000 for each of the 274 alleged infringements, a figure that could push total exposure well above $40 million if the court awards the maximum per-work amount. The filing was submitted by Authentic Brands Group, the entity that controls the Elvis brand, together with several other parties that share ownership of the singer’s publishing and trademark assets.
Anthropic’s AI training practices and prior music litigation
The complaint points to the behavior of Claude, a text-generation model that does not produce audio but can draft lyrics, verses, and even describe musical hooks. According to the allegations, Claude has repeatedly generated outputs that echo the structure, hooks, and phrasing of the copyrighted songs, even after Anthropic reportedly added “guardrails” intended to prevent such copying. In a recent internal test, the model allegedly replied that its renditions were “very close to the originals: same structure, same hooks reused with light rewording, and full sets of verses/choruses/bridges,” suggesting a drift from mere inspiration to direct replication.
Anthropic is not new to music-related lawsuits. Earlier in the year, the company faced suits from major publishers—including Universal Music Publishing, Concord, and BMG—over the alleged use of hundreds of thousands of lyrics to train Claude. In August, a separate action by Round Hill and Suno alleged mass infringement of their publishing catalogs. Those cases collectively involve claims over 500 works, 853 lyric infringements, and more than 20,000 songs, with potential damages ranging into the billions of dollars.
Broader implications for AI developers and creators
The Elvis case stands out because of the cultural weight of the name involved; jurors instantly recognize the king of rock ’n’ roll, even if they are less familiar with the behind-the-scenes publishing entities. While the monetary exposure per song is quantifiable, the broader question remains unsettled: does the act of feeding copyrighted material into a machine-learning model constitute fair use, or is it an unlawful appropriation of protected expression? A 2025 ruling on a separate Anthropic case involving books suggested that training on copyrighted text might be deemed fair use, but storing the works in a permanent repository was not. That precedent could inform the upcoming arguments in the Elvis suit.
For developers of AI tools that generate music-related content, the lawsuit underscores the importance of transparency about training data sources. Companies are being urged to keep detailed licensing records and to clarify in their terms of service whether copyrighted works were used. Creators who rely on AI for songwriting, scoring, or captioning should anticipate stricter compliance checks and possibly higher licensing fees if courts begin to favor rights-holder arguments.
At the time of publication, Anthropic has not responded to inquiries. The next procedural milestone is likely a discovery phase where both sides will exchange evidence about how Claude was trained and what safeguards were in place. The outcome could set a precedent that shapes the economics of AI-driven creative tools for years to come.
