UMG and ElevenLabs Build a Licensed AI Music Platform on Opt-In Artist Catalogues
Universal Music Group and ElevenLabs have signed a multi-year licensing agreement that will produce a licensed AI music platform for fans, alongside a wider collaboration to build AI audio products for artists and songwriters. The two companies confirmed the deal on September 10, 2026. It is the first major-label agreement for ElevenLabs, an AI audio company valued at $11 billion, and it reverses the stance UMG held for years as one of the music industry's most aggressive litigants against unlicensed generative audio tools.
The platform is not finished. UMG and ElevenLabs describe it as in development, with no launch date, pricing model, or final feature list disclosed. What the companies have outlined is a fan-facing product built on recordings from UMG artists who agree to take part. Users would be able to generate authorised remixes and mashups, produce what the two companies call new track interpretations, and create personalised vocal experiences.
| Element | Detail |
|---|---|
| Structure | Multi-year licensing agreement plus wider strategic collaboration |
| First product | Fan-facing AI platform: remixes, mashups, new track interpretations, personalised vocal experiences |
| Artist participation | Opt-in; UMG artists must choose to be included |
| Second workstream | Joint AI audio products for artists and songwriters |
| Rights management | Integration with AI audio models for per-use tracking |
| Status | In development; no launch date, pricing, or feature list announced |
Inside the Licensed AI Music Platform
Artist consent is the load-bearing element of the design. Participation is optional, so only opted-in recordings would feed the platform. That leaves the available catalogue smaller than UMG's full library and more expensive to administer, because rights have to be cleared and tracked artist by artist instead of covered by a single blanket grant.
Legal certainty is what the label buys in return. An opt-in catalogue removes the infringement exposure that hangs over generative audio products trained on scraped recordings, and it gives UMG a defensible answer to artists who object to AI uses of their work. It also creates a template: terms can be renegotiated artist by artist and release by release as the technology and the market change.
Rights management sits in the technical brief rather than being attached afterwards. The collaboration includes integration of rights management with AI audio models, which points to per-use tracking of which recordings are used and how often, so that payments can be routed back to rights holders. Without that plumbing, a generative product built on licensed music cannot scale, because nobody can bill or pay accurately at volume.
Opt-in also redistributes leverage toward the performers themselves. An artist who declines is not exposed to synthetic versions of their voice, and an artist who joins gains a contractual say over which uses are permitted. The cost is fragmentation: the platform's appeal to fans depends on how many marquee names sign, so UMG has to sell participation internally before it can sell the product externally.
Voice is the sharper edge of the feature list. Personalised vocal experiences imply synthesis or manipulation of identifiable voices, the use case that has drawn the loudest objections from performers. Making that opt-in and contract-bound addresses the objection at its source, but it also means the platform's most distinctive capability will be available only for artists who actively agree, at least at launch.
None of the economics can be modelled yet, because the two companies have not said whether fans will pay a subscription, buy generation credits, or create for free with rights holders paid from a pooled royalty. Each option produces a different split between UMG, ElevenLabs and the artist, and per-use tracking only becomes commercially meaningful once one of them is chosen. A credit model would additionally force the two partners to agree on how a single remix is valued before either side can quote artists a rate.
The second workstream runs in the other direction. UMG and ElevenLabs will jointly develop AI audio products aimed at artists and songwriters themselves, not only at fans. That addresses a different commercial problem: giving professional creators tools that work on their own material, with rights held by them. Sir Lucian Grainge, UMG's chairman and chief executive, has framed the opportunity in those terms, arguing that the most promising uses of AI in music are the ones that put artists, songwriters and fans at the centre of the product.
From Courtroom to Clearing House
UMG's position has shifted sharply. Through 2023 to 2025 the label was among the most aggressive litigants against unlicensed generative audio tools, treating scraped training data as a direct threat to its core asset. The ElevenLabs agreement converts that adversarial stance into a supply relationship. The catalogue that was the object of litigation becomes the input to a licensed product, and UMG becomes a licensor with a stake in the output.
The incentives are not subtle. Litigation can stop a competitor but cannot collect revenue from fan demand for AI-assisted remixes. Licensing captures that demand and sets the terms under which it is served, which is a stronger position than waiting for rulings that could go either way. Every remix that runs through an authorised platform is a transaction UMG can see, price, and charge for.
The shift is not confined to one label or one startup. Generative audio moved from novelty to licensing negotiation in roughly two years, and the terms being set now will govern how much of the value in AI music flows back to rights holders. Labels hold the catalogues that make fan-facing remix products worth using, which gives them pricing power that model builders do not have on their own.
ElevenLabs brings the other half of the equation. Its $11 billion valuation is more than twice that of Suno, a rival in AI music generation, and that capital position matters because licensing at catalogue scale is expensive. Rival major Warner Music Group has separately moved toward a partnership with Suno, which suggests the majors are converging on the same conclusion: a licensed, opt-in catalogue paired with per-use rights management is worth more than an injunction.
For unlicensed rivals, the agreement raises the cost of staying outside the system. Suno and similar services built on scraped audio carry continuing legal exposure and now compete against an alternative that arrives with major-label catalogue attached. If the UMG platform shows that fans will pay for authorised remixes, the argument that infringement was the only route to market loses its commercial cover.
Whoever operates the clearing layer for AI music captures the margin. A platform that can verify which recording was used, clear it, and pay the rights holder in a single pass becomes the chokepoint that every AI-generated remix has to pass through. UMG's catalogue supplies ElevenLabs with content, while ElevenLabs' models and infrastructure give UMG a distribution channel it does not have to build. Financial terms were not disclosed, so how that margin is divided remains unknown.
For anyone building on top of the platform, the rights layer is the part to watch. If UMG and ElevenLabs expose per-use clearance as an API-level service, smaller developers could license the same pipeline instead of training on unlicensed audio and absorbing legal risk. If the clearing capability stays internal, the deal entrenches a two-tier market in which only well-capitalised companies can afford authorised catalogues.
Why this matters
The UMG-ElevenLabs agreement shows that the licensing question in generative audio is being settled by contract rather than by court. Media and AI teams building products on copyrighted material now have a template to negotiate against, and a reminder that the compliance layer, not the model, may be the durable asset. The practical test comes when the licensed AI music platform launches and the first per-use payments reach artists. Until a launch date, a price, or a feature list is published, the terms of that settlement stay unverified.
Related Articles
- ElevenLabs Launches Music v2 with Dynamic Mid-Track Genre Shifting
- Stability AI Series B: Music, Gaming Giants Take Stakes
- Anthropic Copyright Lawsuit Over Lyrics Puts a Price on Training Data
✔Human Verified
Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.