The two AI music generation companies named in the RIAA's June 2024 lawsuits are both bigger than they were when the suits were filed. Suno crossed 47 million monthly active users in March, up from 19 million when the suit was filed, and Udio crossed 12 million monthly users in the same window, up from 4 million. Both companies argue the lawsuits will be resolved on fair use grounds, and both have continued to ship more capable models throughout the litigation period. Suno v4 launched in November 2025 and Udio 1.5 launched in February of this year. The summary judgment hearings are scheduled in the Southern District of New York for July 14, with most legal observers calling the case the most consequential AI copyright matter currently pending.
The technical jump from v3 to v4 in Suno's case is real. The new model handles vocal performance, instrumental texture, and song structure at a quality that, in blind testing by The Verge in December, fooled musicians 41 percent of the time when compared against actual human studio recordings. Udio 1.5 made similar progress on production quality, particularly around drum sounds and stereo imaging. Both platforms now offer a paid tier with commercial licensing for the user generated tracks, priced between $10 and $30 per month, and both have crossed an annualized revenue run rate above $200 million according to estimates from PitchBook and Bloomberg.
The RIAA's case rests on the argument that Suno and Udio trained their models on copyrighted master recordings without licensing them. Both companies have not formally confirmed or denied training on RIAA member catalogs, although court filings have included specific examples of generated audio that closely matches the timbre and chord progression of identifiable Universal, Warner, and Sony recordings. The defense in both cases leans on the precedent set in the Andy Warhol Foundation v. Goldsmith case from 2023, which narrowed the fair use doctrine for transformative use, and the Author's Guild v. Google decision from 2015, which broadly protected machine reading of copyrighted works.
The major label response has shifted from pure litigation to a parallel licensing strategy. Universal Music Group struck a licensing deal with Klay AI in October 2025, then with Stability AI in February of this year. Sony Music partnered with ElevenLabs on a voice cloning service for select label artists in March, beginning with Selena Gomez and ending up at fourteen artists by April. Warner Music Group has been the most cautious, with CEO Robert Kyncl publicly stating that the company would not license catalog to AI training without a clear precedent in the Suno and Udio cases.
Independent artists have been more enthusiastic adopters of AI generation than major label artists, partly because the legal exposure for an independent artist using AI to demo or finish a track is materially lower than for a label signed artist whose contract requires master ownership clarity. A Music Business Worldwide survey of 1,400 independent artists in March found that 47 percent had used a generative AI tool in the past 90 days, most often for demos, beat starts, or instrumental ideas. Spotify reported in its Q1 letter that AI assisted tracks now make up between 6 and 9 percent of monthly new uploads to the platform, up from 2 percent a year earlier.
Streaming platform response has split. Spotify has built fingerprinting tools that flag suspected fully AI generated tracks for human review and removed roughly 80 million tracks in 2025 across various violations. YouTube introduced a Music AI principles statement in November and partnered with Universal on the Dream Track tool that lets approved creators use a small set of artist voice models. Apple Music has remained the most restrictive, requiring all uploaded tracks to be human created and certified by the distribution partner, although the policy is loosely enforced.
For the working musician, the question is no longer whether AI tools belong in the workflow but how to set the boundary. A producer mixing for a major label release still cannot run vocal stems through Suno's stem extraction tool without violating the master license. A bedroom artist working on their own song can run that same stem through any of three or four tools and the only question is whether they want the result. The space between those two scenarios is where the legal and contractual clarity is still being written.
The RIAA case timeline matters because the summary judgment ruling will likely set the framework for the next decade of music creation. A ruling in favor of the labels on training data licensing would force every AI music tool to license catalog or rebuild training corpora from scratch. A ruling on fair use grounds would allow the current trajectory to continue, with the labels pursuing licensing deals as a parallel revenue stream rather than a defensive posture.