A significant federal appeals court ruling on September 29, 2026, has set a landmark precedent against AI companies claiming fair use for training on copyrighted materials. The case involved ROSS Intelligence, which copied Thomson Reuters’ Westlaw legal research content to build a competing AI tool without a license. The court found that copying editorial content to create a product that directly competes in the market causes economic harm and does not qualify as fair use. This decision, although about legal research, directly influences ongoing lawsuits against AI music generator Suno, which trained its model on millions of copyrighted songs without permission.

Suno’s defense focuses on the generative nature of its AI, which creates new music rather than reproducing existing works. This distinction raises the issue of “transformative use,” a key component of fair use. However, a recent ruling by a Munich court in Germany has undercut this argument by finding that Suno’s AI could reproduce songs substantially similar to copyrighted works, effectively memorizing and generating copies. This judgment challenges the claim that AI outputs are wholly original, signaling trouble for Suno as U.S. courts may follow a similar legal interpretation, putting the company’s fair use defense at serious risk.

Despite some licensing deals Suno has made, particularly with Warner Music Group and BMG, major industry players Universal Music Group and Sony Music Entertainment continue their litigation efforts, convinced the AI has unlawfully used their catalogs. Additionally, Round Hill Music has sued Suno seeking over $1 billion in damages, while a class action by musicians including Jason Isbell alleges that Suno’s AI encodes artist identities without consent. These ongoing lawsuits highlight that many major rights holders view settlements as business arrangements rather than resolutions of the fundamental legal questions concerning AI training and copyright.

For creators and rights holders, the ROSS Intelligence ruling marks a critical judicial trend favoring copyright holders over AI companies asserting fair use. The key test will be forthcoming court decisions, such as the expected summary judgment by Chief Judge F. Dennis Saylor IV in Massachusetts, which could define whether training generative AI models on copyrighted recordings constitutes infringement. This legal battle will shape the future of AI in music, influencing rights protection and licensing in an era where AI-generated content grows increasingly sophisticated and prevalent.