Federal Appeals Court Rejects AI Training Fair Use Claim, Raising Stakes for Music-AI Lawsuits

On September 29, 2026, a federal appeals court issued the first appellate decision on whether training artificial-intelligence systems on copyrighted material qualifies as fair use. The court rejected the defense, emphasizing that copying the content to build a competing product caused market harm. Although the case involved ROSS Intelligence’s use of Westlaw editorial summaries, the judgment is being treated as a benchmark for all AI-training disputes, including those in the music sector.

Forbes

ROSS Intelligence sought to create a rival legal-research tool by copying Westlaw’s editorial content and arguing that the copying was merely “learning.” The trial court dismissed that fair-use argument, and the appellate court affirmed, noting that the competitor directly threatened Westlaw’s market. Suno’s lawyers point to a key distinction: ROSS did not generate new works, whereas Suno’s platform creates original-sounding songs. Nonetheless, the appellate decision underscores that market competition is a decisive factor in fair-use analyses.

Forbes

Suno has entered licensing agreements with Warner Music Group and BMG, but two major labels—Universal Music Group and Sony Music Entertainment—continue lawsuits in the United States. Those suits allege that Suno trained on millions of recordings and seek to add over 61,000 copyrighted tracks to the complaint. Separate claims accuse Suno of ripping audio from YouTube in violation of the Digital Millennium Copyright Act. Round Hill Music filed a federal suit in California in August 2026, seeking damages that could exceed $1 billion and stating it will not settle. In September 2026, a group of musicians led by Jason Isbell launched a class action in Massachusetts, alleging violations of right-of-publicity by encoding artists’ identities into the model.

Forbes

Suno maintains that its training practices fall under fair use and that its filters are designed to prevent the generation of recognizable songs. The record labels and rights organizations counter that the AI’s output can closely replicate copyrighted recordings, arguing that the “generative” defense collapses when the model reproduces substantial portions of existing works.

Forbes

The next pivotal moment will be the summary-judgment hearing in the Suno case, scheduled before Chief Judge F. Dennis Saylor IV in the District of Massachusetts. That ruling could determine, as a matter of law, whether unlicensed training on copyrighted music constitutes infringement, setting a decisive precedent for the future of AI-generated music.

Forbes