Universal Music Group is developing an artificial intelligence-based platform with ElevenLabs dedicated to transforming songs from its catalog. The stated goal is to enable users to create remixes, mashups, and new versions of licensed tracks within an environment built through a multi-year agreement between the major record label and the company known for its voice and music generation models.
The most notable detail involves artist control: participation in the future platform will be optional. Therefore, not all of Universal's repertoire will automatically be included in the product and, at least according to the available information, access to tracks will depend on the decisions of the artists and rights holders involved. This marks a substantial difference from the unauthorized circulation of synthetic voices, imitations, and tracks trained or generated without a clear contractual mandate.
Universal has not yet announced the name of the service, its launch date, the territories where it will be available, or the precise form the creative tools will take. Crucial details for the audience also remain to be clarified: whether the results can be shared outside the platform, under what limitations, how credits will be attributed, and how any commercial uses will be handled. The announcement sets a direction, not yet a product with full specifications.
A music catalog as authorized infrastructure
The agreement stems from a premise that is increasingly central to the generative music market: the value lies not just in a model capable of modifying audio or producing new sequences, but in the ability to use recognizable material without sparking an immediate conflict over rights. Universal brings its repertoire, artist relationships, and licensing expertise to the table; ElevenLabs provides its technology for AI-generated voice and music.
Within this framework, a mashup or reinterpretation is not treated as a simple user-generated file, but as content requiring upfront rules: which tracks are available, which transformations are permitted, who can distribute the result, and what compensation is due to the parties involved. ElevenLabs CEO Mati Staniszewski pointed to fair remuneration for artists and songwriters as one of the partnership's key goals. However, the agreement does not disclose either the economic model or the revenue-sharing terms.
The opt-in approach can also serve as reputational protection. For some musicians, allowing fans and creators to rework a song can become an extension of promotion and community engagement. For others, the same mechanism may conflict with their artistic identity, existing publishing deals, or the desire to keep a track untouched. Leaving the decision to those directly involved does not eliminate all tensions, but it avoids framing AI as an automatic entitlement to the catalogue.
Universal accelerates push into regulated generative music
The platform with ElevenLabs is another piece of Universal Music Group's music AI strategy. The company is also working on a similar product with Udio and has already secured licensing deals in the sector with Spotify, NVIDIA, and Klay. The common thread is the idea of shifting part of generative innovation into negotiated agreements, rather than letting catalogues and vocal likenesses end up in models or services without agreed-upon rules.
The timing is significant for the rest of the industry as well. This week, Suno announced its first music model trained on licensed songs obtained from Warner Music Group, BMG, and other industry partners. Following an initial phase marked by fierce clashes between record labels and AI companies, the market is beginning to carve out an alternative: not just challenging the use of copyrighted works, but building products based directly on contractual agreements.
This does not mean the rights issue is settled. An authorized system must establish operational boundaries that are far more granular than those of a standard streaming catalog. Using a recording as the basis for a remix is one thing; recreating a voice, altering lyrics, associating an artist with unwanted messaging, or making a new synthetic version indistinguishable from an original is another. The rules that Universal and ElevenLabs adopt on these points will likely be more important than the software's sheer ability to produce a convincing result.
Fan creativity, with boundaries yet to be defined
For users, the promise is moving from passive listening to the guided reinterpretation of well-known music. The concept of amateur remixing has existed for decades, but AI tools can significantly lower the technical threshold: modifying a song's structure, combining elements from multiple tracks, or generating a new rendition may no longer require traditional studio production skills. It is precisely this accessibility that makes a clear licensing framework necessary, as it dramatically expands the potential volume of derivative content.
The platform will also need to strike a balance between creative freedom and catalog protection. Overly rigid restrictions risk turning it into an unrewarding experience for fans; excessive permissiveness could generate content that artists would rather not see linked to their name. Voluntary opt-ins may be accompanied by tiered options across specific tracks, territories, or use cases, but no details have been provided on this front so far.
Universal and ElevenLabs stated that the partnership will also encompass the development of other products and new fan experiences over the coming months and years. While this phrasing leaves the door open to further applications, it does not yet shed light on their exact nature or functionality. Moreover, the new platform will remain separate from ElevenLabs' Music API and ElevenMusic generator, meaning it should not be viewed as a mere update to the company's existing tools.
For Universal, the bet lies in demonstrating that generative music can be integrated into the music business without reducing artists and songwriters to raw material for models. For ElevenLabs, the deal provides regulated access to one of the world's most important catalogs and a use case distinct from generic tools. Success will depend less on the announcement and more on concrete conditions: licensing transparency, the ability to prevent abuse, verifiable compensation, and a real ability for artists to decide how, and whether, their works enter the experiment.
