For Isbell and his fellow plaintiffs, the issue is not simply whether AI learned from songs, but whether it can commercially mimic the people who made them. Suno, meanwhile, has defended its broader training practices as lawful fair use in separate copyright disputes.

Suno, founded in 2022, built a leading position in AI music by letting users turn short text prompts into finished tracks within seconds or minutes. Its rapid rise — investors have valued it at $5.4 billion — has also made it a central target in the music industry’s fight over how generative models are trained.¹

Before this week, the company was already facing lawsuits from record labels and publishers alleging that it trained on millions of copyrighted songs without permission. Suno has argued in court that such training can be protected by the fair-use doctrine, a defence available in limited copyright cases.¹

On Monday, Jason Isbell, Guy Forsyth, Eduardo Calle and David Lowery took a different route in U.S. District Court in Massachusetts. Seeking class-action status, they accused Suno not of conventional copyright infringement, but of exploiting the distinctive identities behind the music: their voices and styles.¹

The complaint alleges that Suno “encoded musicians’ identities into its A.I. model to create its commercial product without consent,” then used their names for its own commercial benefit.¹ That framing could matter: it asks whether a model’s output can unlawfully appropriate an artist even when the dispute is not centred on ownership of a particular song.

The plaintiffs point to what they say is a practical test of that claim. Although Suno says users are blocked from prompting with specific artists’ names, the suit alleges the system accepted “jason isbell” and generated “Paper Bell,” an Americana-style track said to echo Isbell’s clear vocals and country twang.² The case now puts the boundary between musical influence and AI identity imitation squarely before the court.