TIDAL stopped paying the machines

On 29 June, TIDAL announced a rule that took effect on 15 July and that no major streaming service had adopted before: tracks made one hundred percent by artificial intelligence no longer collect royalties. Not pulled from the catalogue. Not hidden from search. Simply unpaid — still playing exactly as they did yesterday, minus the wallet.

The line is drawn at how much of the track a human touched. AI tools remain allowed — provided a person is using them; the machine may be the pencil, but not the author. Fully automated production loses its eligibility. Qualifying releases are picked out by automated detection and marked with an “AI” badge in the interface.

A flagged track loses three things: its place in the royalty pool, access to direct-to-fan sales, and — if it impersonates a real artist or carries misleading metadata — its place on the service entirely. Everything else keeps playing. Which creates a category that did not exist before: music you are permitted to hear but not to pay for.

Other platforms are still working on visibility rather than the wallet. Spotify labels tracks and filters spam; Apple Music and Deezer lean on detection tools that flag AI songs inside playlists. TIDAL is the first to say out loud that this is not the same product. The AI music market is projected at four billion dollars by 2028 — enough money that the decision is more than a statement.

The interesting part is not the ban but the border. Somebody has to decide where the human ends inside a piece of music, and in this case that somebody is an algorithm auditing another algorithm. For the first time, the question of whether it sounds good has been formally separated from the question of who stands behind it. In a room with a piano in it, the second question answers itself: a person is sitting there, and they get paid.