Suno v6 Is Trained on Licensed Music. What That Means for AI Musicians

For most of its short life, AI music generation has carried an asterisk. The tools were impressive, but nobody outside the companies could say exactly what the models had learned from, and major labels were in court arguing that the answer was their catalogs, used without permission. In September, one of the biggest names in the space changed its pitch.
TechCrunch reported on September 9, 2026 that Suno had launched a new model family, Suno v6, which the company says was developed using licensed data from Warner Music Group, BMG and Believe. Suno also said v6 was not trained on the data used for its earlier models, and that the previous models are being retired as v6 rolls out. If you make music with Suno, or with any generator, this is worth understanding properly rather than as a headline.
What Actually Launched
According to Suno's announcement and TechCrunch's coverage, v6 is three models rather than one. The base v6 model is aimed at controlled, steerable output and is available to paying subscribers. A second model, v6-wild, is pitched as more experimental and is also for paying users. A faster v6-mini is available to everyone, including free accounts.
The licensing story has a history. Warner Music Group settled its lawsuit against Suno in November 2025 and agreed to work with the company on licensed models, which TechCrunch covered at the time. Not every rights holder has followed. As of TechCrunch's September report, Universal Music Group, Sony and the German collecting society GEMA were still pursuing legal action against the company. Suno has also not published a full list of what went into v6 training, so "licensed" here means "licensed from the partners named," not a complete public inventory.
What Licensed Training Changes for You
The most practical change is in the conversation around your work. A large part of the criticism aimed at AI musicians has been about where the model's knowledge came from. A model built on licensed catalogs, with labels and distributors publicly involved, gives you a cleaner answer when a listener, a collaborator or a platform asks how your track was made.
It may also matter for distribution over time. Streaming services are adding labels and filters for AI music, and the industry is clearly separating tools built with rights holders from tools built without them. Nobody can promise how platforms will treat each category, but working with a tool that has public licensing deals is unlikely to count against you.
What It Does Not Change
Licensed training data is about what went into the model. It does not automatically settle who owns what comes out. Your rights in a generated track are set by the tool's terms of service for your plan, and by copyright law where you live, which in many places still requires meaningful human authorship before a work is protected. Read the current terms rather than assuming.
- Licensed inputs do not mean you can release a song that imitates a specific living artist's voice or style without consequences.
- Free and paid tiers have historically come with different commercial terms. Check which tier you were on when you made a track.
- Ongoing lawsuits concern earlier models. If your back catalog was made on retired versions, keep records of when and how each track was made.
- A model being licensed does not make undisclosed AI use acceptable to platforms that ask you to declare it.
A Short Checklist for Suno Users Right Now
Model changes are a good moment to tidy your records. A few habits will save you trouble later, whatever happens in court or in platform policy.
- Note the model version, date and plan tier for every track you release, especially anything made before v6.
- Save your prompts, lyric drafts and any stems you edited, so you can show your own contribution.
- Re-read the terms of service now, since a new model family is often paired with updated terms.
- Expect your old prompts to behave differently on v6 and budget time to re-learn the tool.
- Keep exported masters of finished work on your own storage, not only in the generator's library.
The Bigger Picture for AI Music
Suno's shift is part of a broader pattern. Rights holders are moving from suing AI music companies to licensing them, at least where they can agree terms, and platforms are building ways to label and sort AI music. For creators, the direction is toward more transparency, not less. The musicians who will do best are the ones who can explain their process plainly: which tools they used, what they wrote or performed themselves, and what they own.
That is not a reason for anxiety. It is a reason to treat your workflow the way a working producer treats session paperwork. The music is the point, but the records protect it.
It is also worth remembering that this is one company's approach. Other music generators have their own training data, licensing deals and terms, and those can change too. Whichever tool you use, the same questions apply: what is it trained on, what do the terms let you do with output, and can you show your own contribution.
Sharing Your v6 Work
If you are experimenting with the new models, Kelyx is a free place to publish AI-made music alongside the process behind it. Sparks can describe the tools and steps they used right on the track page, and listeners who like what they hear can leave a charge and follow along. It is a community built around AI work, so you do not have to apologize for the method.
Frequently asked questions
Was Suno v6 trained on licensed music?
According to Suno and TechCrunch's September 9, 2026 report, the v6 model family was developed using licensed data from Warner Music Group, BMG and Believe, and was not trained on the data used for earlier Suno models. Suno has not published a complete list of the training material.
Do I own songs I make with Suno v6?
Ownership depends on Suno's terms of service for your plan and on copyright law where you live, not on how the model was trained. Licensed training data does not by itself give you rights in the output. Check the current terms and keep records of your own creative contribution.
What happens to tracks made with older Suno models?
Suno says its previous models are being retired as v6 rolls out. Tracks you already made remain yours under the terms that applied when you made them. Keep notes on the model version, date and plan tier for older releases, and store exported copies yourself.