The AI Music Flood Goes to Court: UMG and Sony Want DistroKid's Upload Numbers

On October 14, lawyers for Universal Music Group and Sony Music walked into federal court in Manhattan with a request that shows how the music industry's war over AI-generated music has shifted. They are no longer only arguing about how AI models are trained. They want hard numbers on how much machine-made music now flows through the distribution pipeline, and they want those numbers from DistroKid, the world's largest independent music distributor.

The motion asks a judge to compel DistroKid to comply with a subpoena sent in September, seeking the company's monthly submission figures and the results of its AI detection tools. The labels say the information is evidence in their long-running copyright lawsuit against Suno, the AI music generator they have accused since 2024 of copying protected recordings to train its systems. Suno disputes the allegations and argues its training is fair use.

The numbers in the filing

The filing puts DistroKid's scale in stark terms. According to the labels, DistroKid delivered nearly 12 million tracks to a single major streaming service between March and September 2026, accounting for more than half of all tracks delivered to that service in that window, and more than all other distributors combined. The filing does not name the service, and DistroKid has not publicly confirmed the figures.

What the labels really want is what DistroKid's own detection technology flags. DistroKid screens incoming submissions for AI generation, and the labels want a month-by-month look at what the screens catch. Suno, by contrast, does not track where its outputs go once they leave its platform, the filing says, which makes the distributor, sitting upstream of the streaming services, the best available window into the real scale of the influx.

The labels call DistroKid one of the primary conduits by which the streaming market is being flooded with AI-generated recordings. In the motion, their attorneys wrote that DistroKid's AI-detection results are direct evidence of a flood driven by what they describe as Suno's industrial-scale copying.

The lawsuit that began as a question of how AI music is made has become a question of how much of it exists.

A two-front legal war

The compulsion motion connects to a second battle the labels opened last month. On September 15, UMG filed a 52-page complaint against DistroKid itself in federal court in Delaware, accusing the distributor of copyright infringement and deceptive trade practices. The complaint alleges DistroKid delivered mass quantities of what UMG describes as AI-generated, SEO-optimized, infringing tracks to streaming platforms, alongside recordings that speed up, remix, or sample existing works without authorization.

That complaint names 1,000 UMG recordings the label says were copied without authorization, seeking up to $150,000 per work in statutory damages, a theoretical ceiling of $150 million. It also asks the court to order DistroKid to remove all infringing tracks and terminate the accounts of repeat infringers. Notably, a footnote says the suit does not challenge AI-generated music that is clearly disclosed as such.

The labels say they need data from both fronts. DistroKid disputes the relevance of its upload figures, arguing that the volume of AI-generated music on streaming platforms says little about whether Suno copied copyrighted recordings to train its models. The labels counter that their case also covers the commercial consequences of that alleged copying, including the displacement of human-made music, which goes directly to fair use. Under federal copyright law, one factor courts weigh in fair use is the effect of the challenged use on the market for, or the value of, the original works.

Why the numbers are hard to trust

Even if a judge orders DistroKid to hand over its detection results, the labels' own account acknowledges a gap: a count of uploads alone would not establish that those recordings came from Suno, or that Suno's conduct caused specific financial losses. The number of AI uploads measures the size of the flood, not its source.

DistroKid's own enforcement has already drawn pushback from the people the platform is supposed to serve. After the September lawsuit, the company said it had removed a very small number of recordings, a move some independent musicians challenged, saying legitimate work was taken down without adequate explanation. Amanda Ferri, DistroKid's vice president of artist services, said the company strongly disputes the allegations underlying UMG's lawsuit and regrets the disruption to affected artists, adding that it is working to minimize disruption while protecting artists' ability to release their music.

The scale of the problem is visible elsewhere in the industry. Deezer disclosed in July that it receives nearly 90,000 fully AI-generated tracks every day, and that on peak days such tracks account for more than half of all new deliveries. Yet they generate only 1 to 3 percent of streams, and the company estimates that up to 85 percent of those streams were fraudulent in 2025. This month Deezer announced it would let listeners filter AI-generated music out of search results entirely.

What it means for listeners

The outcome of the compulsion motion will not settle whether AI-generated music belongs on streaming platforms. But it could set the terms of how the music industry counts it. If the court orders DistroKid to produce its detection data, the labels gain a quantified story of displacement to take into the fair-use fight. If it refuses, the industry's central claim, that a flood of machine-made tracks is drowning human artists, stays a metaphor rather than a measurement.

For listeners, the practical stakes are already visible: recommendation feeds and search results shaped by a catalog in which the majority of new arrivals may never have been made by a person. Streaming services have responded with a patchwork of policies. Bandcamp banned AI content, Tidal restricted its monetization, Spotify introduced disclosure measures, and Deezer is moving toward an opt-out. Whatever the Manhattan court decides about a subpoena, the question of who gets to measure the flood, and who believes the measurements, is now part of the culture of music itself.