The largest music companies are making deals with the biggest AI music services, underscoring that we are no longer in a paradigm where the music economy is a world of files, works, and recordings that can be released, measured, and settled as fixed units. We are rapidly moving into a reality where music is no longer just finished recordings and works, but can exist as something that constantly mutates and only briefly takes form in the moment we ask for it.

In the “old” world, music was bound to something that could be fixed. A recording, a file, a composition, a lyric. Only when it could be released as a static unit could it be registered, credited, tracked, and settled. The entire rights system is built around that logic. The product must have a fixed and stable form. In a malleable possibility space (latent space), music is not one fixed version, but something that can continuously take many forms. It can be changed, remixed, and reinvented without needing to end as a final release.

This does not mean that liquid music cannot be tracked. It means that tracking can no longer assume the work stands still. You can recognize it, but not lock it down. You can register it, but not as one static file or recording. We can point to style, origin, and sources – but not necessarily to one final version that the economy can be tied to.

This challenges our way of commercializing and settling music. “Per track,” “per stream,” and “per use” only make sense when music is sold in units and objects. When music instead exists in a state of constant possibility – and only takes its temporary form in the moment we ask for it – it becomes gradually harder to tie economics to locked and measurable music units.

In conventional production logic, music moves through fixed roles and phases – from composer to musician, producer, mixer, and master – until it can be released as a locked file. The master. In the new reality, it all melts together into one state that takes new form when we ask for it. In this paradigm, it is less about finishing and more about shaping music for immediate, temporary consumption.

The new AI deals are therefore also largely about defining when a fluid potential becomes commercial exploitation. Majors are opening up for their catalogues to be dissolved into a possibility space, but only in closed systems, so output cannot flow freely out and compete as releases. This protects works and recordings as commercial and cultural objects and ensures that the transition to “liquid music” happens with respect for origin, value, and ownership. The question is how these closed systems will compete with open systems that allow this fluid music content to flow freely into the traditional music economy. We can’t expect these to all be sued into the ground as they continue to pop up in new forms, especially as synthetic music data starts to proliferate and training sources become much more blurry from a legal and technical standpoint.

If music is no longer just units but a fluid space of possibilities owned and controlled by tech companies, we must define who profits both on the platform and when possibility becomes product. People will continue to create fantastic music, probably more than ever, but the right to make a living from it must not slowly disappear into that same fluid possibility space controlled by those who built the infrastructure.

Illustration:
J. M. W. Turner – “The Fighting Temeraire” (1838)