Tidal has decided to stop paying royalties on tracks it identifies as 100 percent AI-generated, and to flag those same tracks for listeners starting July 15. The policy, announced on June 29 by the Block-owned paid music streaming service, leaves AI music playable on the platform while stripping it of monetization, and reserves the right to extend the label to tracks that are "substantially AI-generated" as the company's detection stack improves.
The architecture underneath the announcement is where the real story lives. Tidal is building a royalty firewall on top of a classifier it has not publicly named, and betting that "100 percent AI-generated" is a stable category at a moment when generative tools are rapidly blurring the line between human and machine authorship. The company's stated rationale, in its post on X, is that "Tidal's priority is ensuring royalties go to original works directly produced, written, and performed by people. We will therefore not knowingly attribute royalties to music we identify as wholly AI-generated." That sentence also marks the load-bearing distinction the policy will eventually have to defend: "wholly" AI today, "substantially" AI someday.
According to The Verge's reporting, Tidal is combining its own detection tools with signals from partners and submitters, but has not disclosed which tools those are, what their accuracy looks like, or how artists can appeal a misclassification. The policy treats detection as a solved-enough problem to gate royalties on, while explicitly leaving detection itself evolving. That gap between the policy's confidence and the underlying tooling is where disputes are most likely to surface, because it determines whether the next phase is a clean rollout or a steady trickle of misclassification claims.
The mid-July enforcement layer goes further. Music Business Worldwide reports that Tidal also plans to remove or block AI tracks tied to fraud: deception, high-volume uploads, and unusual streaming patterns. That piece of the policy is a familiar anti-stream-manipulation mechanism dressed up in AI language. What is new is that Tidal is collapsing two distinct problems (fraud and AI provenance) into one enforcement lane, and asking the same opaque classifier to do the sorting.
The competitive context is part of why the precedent matters less than the headline suggests. Tidal is one of the smaller paid streaming services by global subscriber share, with an audiophile and DJ-leaning positioning rather than the mass-market reach of Spotify, Apple Music, YouTube Music, or Amazon Music. As TechCrunch's coverage frames it, the bigger platforms have so far taken softer tacks: Spotify has leaned on artist verification and disclosure rather than demonetization, Deezer has signaled it would downrank AI-heavy tracks, and Variety's writeup describes the same industry-wide uncertainty about detection reliability. None of those comparator positions are independently confirmed inside Tidal's announcement; they are how trade and tech press are reading the moment.
The unanswered questions are the ones worth watching. First: what is Tidal actually using to classify tracks, and what happens when a human artist is mislabeled as AI and loses payouts during the appeals gap? Second: when "substantially AI-generated" labeling lands, who draws the line for a song that used a generative vocal tuning pass, or a drum stem produced by a model and then re-performed by a human band? Third, and most durably: even if detection improves, the underlying supply question (whether the training data for these models was licensed or scraped without consent) is not something any streaming platform can resolve by labeling a finished track.
What to watch next is concrete: the July 15 on-platform label rollout, the first public misclassification disputes, and whether any of the larger DSPs follow Tidal into demonetization or keep their distance from the detection bet the smaller service has made without naming it.