TIDAL has updated its AI policy with a change that is simple to state and consequential to live with: music it identifies as fully AI generated will no longer earn royalties on the platform. The demonetization is effective immediately as of June 29, and a separate consumer facing change, an AI badge on wholly AI tracks plus the removal of fraudulent uploads, takes effect on July 15.
The important nuance is that this is not a ban. AI generated music can still exist on TIDAL. It will simply be labeled with an AI tag so listeners know what they are hearing, and it will not collect royalties or be eligible for the direct to fan sales the platform offers. You can upload it, people can play it, but it will not pay. That distinction, allow but do not monetize, is the whole story, because it sidesteps the unwinnable fight over removing content and instead targets the incentive.
This is also a different emphasis from how other streaming services have approached the issue. The earlier wave of responses was largely about fraud: platforms talking about bot driven streams, fake artists, and the share of uploads that turn out to be scams designed to siphon royalties from the pool. Those are real problems, but they frame AI music as primarily a security issue. TIDAL is making a values statement on top of that, saying it wants royalties to flow to works produced, written, and performed by people, and that it will not knowingly attribute payouts to music it deems entirely machine made, fraudulent or not.
The policy applies across the board, covering both ordinary label and distributor uploads and TIDAL's self serve Upload tool aimed at independent artists. It arrives in a year when streaming platforms have visibly diverged on AI music, with some like Bandcamp banning it outright, others choosing to tag and disclose it, and others still saying little. TIDAL's position stakes out a middle path that is becoming more common: tolerate the content, inform the listener, and cut off the cash.
The genuinely hard part is the one the announcement cannot fully resolve, which is detection. Drawing the line between fully AI generated and merely AI assisted is a judgment call, and most modern music production already uses some machine help, from mixing to mastering to stem separation. A policy that pays humans and not machines only works if you can reliably tell which is which, at scale, without punishing artists who used a tool along the way. That classification problem, not the principle, is where this will be tested. But the direction is clear and it matters beyond TIDAL: as AI lowers the cost of generating plausible music to nearly zero, the platforms are quietly deciding that abundance alone earns nothing, and that the money is reserved for the human part.