A second federal Suno lawsuit landed on Friday, 18 September 2026, when Universal Music Group and Sony Music Entertainment filed a fresh complaint in Boston’s District of Massachusetts. As reported by Music Business Worldwide’s Tim Ingham and Variety’s Corbin Bolies the same day, the majors allege that Suno copied 60,202 of their sound recordings without a licence and used that corpus to build the generative models behind its AI music service—including the v6 suite launched around 9 September 2026.
This is not a housekeeping amendment to the June 2024 case. It is a separate action with a larger catalogue, a sharper theory about “cleaned” models, and a damages ceiling that forces the industry to do arithmetic in public.
What Universal and Sony allege about Suno v6 copyright
US copyright statute allows willful-infringement damages of up to $150,000 per work. Applied to 60,202 recordings, that produces a theoretical ceiling of roughly $9 billion—a number the complaint floats as statutory potential, not a guaranteed award. The labels also pursue anti-circumvention claims tied to YouTube download-blocking technology, seeking up to $2,500 per act; stacked across the same catalogue, MBW notes that line could add on the order of $150 million. They ask for an injunction against further use of their recordings and demand a jury trial.
The complaint’s centrepiece is doctrinal and technical at once. Suno’s v6 launch messaging—chief product officer Jack Brody told MBW the model was “trained entirely from scratch” without Universal or Sony data, and that the company partners with Warner Music Group, BMG and Believe—is treated by the plaintiffs as insufficient. Their phrase, quoted widely on 18 September, is that v6 is “the fruit of the same poisoned tree.”
In plain language, the Universal Sony Suno theory runs like this: even if a new model is not fed the original WAV files again, training on outputs, user preference choices and knowledge-distillation signals from earlier allegedly infringing models still carries forward expressive value taken from the plaintiffs’ recordings. Preference data, in Suno’s product design, includes which of two generated tracks a user selects—signals the labels say are judgments about audio that itself came from tainted teachers.
Why a second case after the June 2024 Suno lawsuit
The first Universal/Sony action against Suno was filed in June 2024 and centres on a much smaller illustrative set of works (reporting around the August ruling cited 560 works in that case). The majors later sought to add tens of thousands more titles. On 18 August 2026, Judge F. Dennis Saylor IV refused that expansion, preferring to keep the original case focused on the fair-use fight and noting that additional works could be brought separately. Radio Facts and MBW both frame Friday’s filing as that separate suit.
Discovery in the first case, according to the new complaint as summarised by MBW, used audio fingerprinting (Audible Magic) to identify label recordings inside Suno’s training materials. The labels argue the earlier complaint only showed a “minuscule” sample of what was actually copied.
Suno’s fair-use defence in the first matter—that AI music training on copyrighted recordings can be lawful—remains the industry flashpoint. The new complaint tries to undercut that defence by pointing to Suno’s own licensing deals with other rightsholders as proof that a licensing market exists.
Market harm: Deezer’s July 2026 AI-track flood
To argue present injury, the filing cites Deezer’s July 2026 disclosure that roughly half of new daily uploads were AI-generated—about 90,000 tracks a day. Universal and Sony cast that volume as competition for placement, discovery and royalty pools. Whether a Massachusetts jury accepts platform-wide AI spam as proof of harm from one vendor is a trial question; the citation is now part of the pleading record.
People also ask: did Suno respond to the new lawsuit?
As of the 18 September reporting wave, there was no immediate comment from Suno beyond the earlier public launch statements already on record—Brody’s “from scratch” training claim and the Warner/BMG/Believe partnership narrative. This desk will not invent a corporate rebuttal.
Why Music AI rights matter beyond one docket
Friday’s filing seeds the Sunday Profile Music AI rights cluster: label litigation, model versioning and the gap between “licensed partner” marketing and unlicensed training allegations. For a sibling tech explainer on agentic AI capability and safety framing, see our GPT-6 Astra explained piece—different industry, same question of what “trained from scratch” actually means when prior systems leave data exhaust.
Expect the next fortnight’s coverage to track answer deadlines, any emergency injunction briefing, and whether Warner-aligned messaging collides with Universal/Sony’s poisoned-tree theory in open court.
Sources: Music Business Worldwide (Tim Ingham, 18 Sept 2026); Variety (Corbin Bolies, 18 Sept 2026); Radio Facts (18 Sept 2026). Damages ceilings and catalogue counts follow those reports of the complaint.