The "model laundering" case against Suno could reach AI video models

This week's open video releases lean on a trick that a music lawsuit is asking a court to treat as infringement. The trick is distillation: training a small, fast model to reproduce what a bigger model does. A complaint filed by Universal Music Group and Sony Music against Suno on September 18 targets exactly that step, and the legal theory behind it is the one to watch if you use, fine-tune or resell any model distilled from someone else's.

What the labels allege

The joint complaint, filed in Boston federal court and obtained by Music Business Worldwide, accuses Suno of copying 60,202 of the labels' sound recordings without a licence and using them to build the models behind its music generator. US law allows up to $150,000 per work for willful infringement, which puts the theoretical ceiling on those 60,202 recordings at just over $9 billion; a separate anti-circumvention claim of up to $2,500 per act would add roughly $150 million.

The new allegation is aimed at v6, the model suite Suno launched on September 9, which the company said was built with Warner Music Group, BMG and Believe. Suno's chief product officer, Jack Brody, told MBW at launch that "v6 was trained entirely from scratch, from the ground up," on data that "doesn't include data from Universal or Sony."

The labels argue that does not settle liability. Their complaint states: "Training a 'new' model on the outputs of an infringing model does not eliminate the infringement; it launders it, passing the value of Plaintiffs' expression from the copied recordings into the tainted models, from those models into their outputs, and from those outputs into v6."

How the distillation argument works

Suno generates two tracks per prompt. The preference data the labels are pointing at is Suno's record of which of the two a user picked, and the filing calls that a judgment about audio its earlier models produced: "To date, every such signal has been a judgment about audio outputs generated using Suno's unlicensed models."

The complaint describes knowledge distillation as a process "whereby a new 'student' model is trained to reproduce the learned behavior of a predecessor 'teacher' model," and names v4, v4.5, v4.5+, v5, v4.5-all and v5.5 as built on the same copied corpus. In the labels' account, "every model Suno has ever released, from its earliest versions through v6, is the product of, and continues to exploit, Suno's unauthorized copying..."

Suno's answer is that the claim is wrong on the facts and the law. In a statement Suno gave to Engadget, reported by Android Headlines, the company said v6 "was trained on content licensed from our partners, interactions including creations and preference signals from our community, and the accumulated learnings from our team," and that the new claims are "fundamentally flawed on both the facts and the law." That wording puts user creations alongside preference signals rather than inside them.

Why video is the place this lands next

Distillation is not a music-industry technique. It is a standard way to make a slow model fast: a large teacher is compressed into a student that renders in a handful of steps. The five few-step releases SLOP TV covered on September 29 and the explainer of what those releases actually do are both about students distilled from teachers, and the licence on a student model says nothing about where its teacher's training data came from.

That is the gap the Suno complaint drives into. If a court accepts that a student inherits the liability of its teacher, the question for anyone shipping a video model becomes the provenance of the teacher, not the licence text on the student. If a court rejects it, "trained from scratch" becomes a clean-slate argument any lab can make without showing its data.

The first test is a procedural one rather than a verdict. Fact discovery in the labels' original 2024 case against Suno closes on September 30, and the new complaint adds its 60,202 recordings to a fight the labels had tried to fold into that case before Judge F. Dennis Saylor IV refused on August 18, saying the additions would delay a trial he wants decided on fair use. Suno admitted in a September 1 filing that it obtained audio from YouTube using the downloader YT-DLP.

The new case is Universal Music Group and Sony Music Entertainment v. Suno, filed September 18 in the District of Massachusetts. Discovery in the original 2024 case closes on September 30.