Tidal moves to label AI-made music and pull royalties — a small step in a much larger fight

On 29 June 2026, music streaming service Tidal announced that tracks uploaded to its platform which are "wholly or substantially created using AI" must be labelled as such, and that songs meeting that threshold will not be eligible for royalty payouts. The policy, reported first by Variety, is the most concrete rule any major digital service provider has yet posted on the question that has been nagging the recorded-music industry for two years: what does a streaming economy owe a song no human made?

The answer Tidal has reached is provisional, narrow, and almost certainly temporary. It treats AI as a labelling problem and a payments problem — not, for now, as a copyright problem or a market-access problem. That framing is the first tell about where the industry's centre of gravity sits, and where it does not.

What the rule actually covers

The threshold Tidal has set is "wholly or substantially" AI-generated. That language leaves a substantial middle ground untouched: tracks where AI is a co-writer, a producer, or a vocal stand-in remain inside the royalty system. The rule, in other words, draws its line at the song with no human author at all — a useful surrogate for "is this protectable expression?" but a poor surrogate for "did anyone get paid for this work?" The distinction will not survive contact with the first dispute over an AI-assisted hit.

Reporting on the announcement does not yet specify what verification mechanism Tidal intends to deploy. Artists self-certifying at upload is the obvious default and the obvious failure mode: there is no public indication Tidal is deploying audio-forensic detection, watermarking, or third-party provenance tools at the gate. The company has framed the policy as a transparency measure; whether it functions as one will depend on enforcement details that have not been disclosed.

Why the music industry got here

The pressure on platforms to act has been building since 2023, when synthetic vocals and fully generated tracks began circulating on DSPs alongside human-recorded music and began quietly accruing royalties. Rights organisations — among them the major-label-affiliated collecting societies and the larger independent bodies — have spent the intervening period arguing about whether a model can be an author, whether a prompt can be a composition, and whether training data is a licence or a taking. None of those questions are settled in court. What is settled is that streams of unattributed AI tracks were being paid out, and that the labels representing the human writers and performers being undercut wanted that money redirected.

Tidal's move lets the company claim it is responding to that pressure without having to take a position on any of the underlying copyright fights. It is, in effect, a deferral with a label attached.

The counter-narrative

The case against labelling or withholding royalties is straightforward: AI tools are now part of how records are made. Vocal tuning, drum replacement, mastering chains and pitch correction have all been machine-augmented for years. A regime that pays on the basis of how much human muscle touched a session risks freezing the production technology in place at the moment the platforms decided to ask the question. Independent artists who have built workflows around AI co-writing tools have a real stake in not being pushed off monetisation rails entirely.

There is a second critique from the other side. Major labels, which have themselves been striking licensing deals with AI model providers, benefit from any rule that formalises the boundary between "AI" and "human-assisted" — because they can credibly sit on the human side of that line in a way open-source community models and bedroom producers cannot. A royalty regime built on disclosure of AI use is also, inevitably, a regime built on the documentary infrastructure of the labels that already know how to file the paperwork.

The structural frame

The recorded-music industry has spent two decades watching its product be aggregated, intermediated, and priced by platforms it does not own. The AI fight is, among other things, a fight about whether the next generation of production tools will be subject to the same intermediation — or whether the platforms will simply absorb those tools, the way they absorbed the distribution of records in the 2000s. Tidal's labelling rule is a small administrative act. The question underneath it is whether the streaming economy will pay for songs at all, or whether it will pay for verified human labour, with the difference accruing to whichever side of that line has the better paperwork.

A counter-position worth taking seriously: the rule may not be about exclusion at all. It may be about price discrimination. A track labelled AI can be priced, recommended, or de-prioritised without being banned; an unlabelled AI track that turns out to be AI can be re-monetised in a different bucket. The honest reading of the policy may be that Tidal is segmenting its catalogue, not restricting it — and that the royalty-withholding component is a stopgap until the segmentation is operational.

Stakes

For artists who use AI as part of their workflow, the rule introduces a disclosure obligation whose downstream effects on payouts, playlist placement, and label signings are not yet visible. For labels and publishers, the rule formalises a competitive moat: human-authenticated catalogues become a more clearly defined asset class. For the streaming economy as a whole, Tidal has now set the pace for disclosure, and the larger services — Spotify, Apple Music, YouTube Music, Amazon Music — will be pressed to either match, exceed, or explain why they have not.

What remains genuinely uncertain is enforcement. The sources do not specify how Tidal intends to detect undisclosed AI use, whether disputed uploads will be reviewed by humans, or how the threshold of "substantially AI" will be operationalised against hybrid productions. Until those mechanics are public, the rule is more declaration than regime — and the industry fight that produces the answer is still ahead of the platform.

— Desk note: Variety broke the policy on the day. Monexus frames this as the start of a labelling regime rather than a copyright ruling, and flags the open question of detection mechanics that the wire reporting has not yet addressed.