More than 80% of United States consumers in research by Universal Music Group (UMG) want a way to exclude fully artificial intelligence (AI) generated music from streaming. For an independent label, the commercial question is where a recording would sit if a streaming service offered that choice: inside the listener’s available catalogue or outside it.
A filter acts before a play
A consumer filter would operate at the point of selection. A listener who excludes a category cannot choose a recording they do not see. That is different from a change to the royalty terms for a recording that does get played. The distinction matters because a catalogue owner could have the same rights and payment arrangements while reaching fewer listeners who use the control.
The UMG finding concerns demand for a way to exclude fully AI-generated music. It does not define how a streaming service would classify a recording or say that a service has introduced such a filter. That is the one implementation question that changes a label’s immediate decision: a classification rule determines which recordings might be affected.
“Fully AI-generated” also sets a narrower question than whether AI was used somewhere in the work. A label deciding how to present its catalogue should identify the recordings it could clearly describe as fully generated, then keep the basis for that description with its release information. For recordings that do not fit a clear description, the useful next step is to ask the distributor what information a service would require under an actual policy.
Classification and permission are separate
A filter is a choice offered to a listener. Permission to use music concerns the rights attached to the recording and the underlying work. One does not settle the other. If a recording falls into a filtered category, its owner still needs to know who controls its rights. If a recording remains available to a listener, availability alone does not answer whether any AI-related input was licensed.
That separation helps an independent label avoid treating a platform label as a rights certificate. The label can document how a recording was made, who authorized material used in it, and who can answer a classification query. Those records support different decisions: the production record supports a description of the music, while the rights record supports an account of what the label may distribute and license.
The practical handoff is to the distributor. Ask which party can correct a service’s classification, what evidence that party needs, and how a correction would reach each service carrying the recording. A label should make that process part of its catalogue records before it has a disputed entry to resolve.
Demand is not a revenue forecast
The more than 80% consumer figure measures interest in a choice. It is not a count of listeners who would switch on a filter, a count of affected recordings, or a royalty rate. Multiplying that percentage by a catalogue’s current revenue would therefore produce a number with no usable payment meaning.
The same measurement problem appears when broad music-market figures are used as shortcuts for catalogue income. An EY–IPRS study estimates that India’s wider music economy exceeds label and publishing revenue by more than three times, but those figures measure different things and do not yield a catalogue’s value per stream. A consumer preference figure is further removed from a royalty calculation: it says what respondents want available, while a label needs to know what happened to its recordings.
Start with the narrowest observable result. If a service offers an AI filter, check which catalogue entries it places in the category. Then compare listening for those entries after the control becomes available, using the service information the distributor can provide. Finally, reconcile any change in receipts against the party responsible for paying the recording royalty. A combined play count can track listening through a catalogue transfer, but it cannot identify which distributor owes the corresponding payment. The same discipline applies here: a change in visibility, a change in plays and a change in money are three separate observations.
Make the catalogue decision recordable
An independent label can prepare a short record for each recording likely to raise a classification question:
- Description: State whether the label considers the recording fully AI-generated and keep the production basis for that statement.
- Rights: Record who can authorize distribution and answer questions about material used to make the recording.
- Delivery: Identify the distributor contact and the route for correcting a service classification.
- Measurement: Keep the recording’s service-level listening and payment records together so a later change can be checked against receipts.
The decision for a catalogue owner is concrete: establish which recordings could enter the category before assigning a revenue estimate to the filter. Consumer demand for exclusion is already measurable; the label’s own classification and payment records determine whether that demand can be translated into an action on its catalogue.