Tidal will host AI music and tag it — but pay it zero royalties

Tidal built its brand on hi-fi audio and paying artists more than rivals. Its new AI policy keeps the AI tracks, slaps an 'AI' badge on the wholly-generated ones, and refuses to send any of them a cent.

In brief

Subscribers describe feeds polluted with releases that borrow a beloved artist's name but clearly aren't them. Tidal's answer: an 'AI' badge on music it detects as wholly AI-generated, automatic removal of impersonations, and a hard rule that 100% AI tracks earn no royalties and no direct-to-fan sales.

One reader cut through it: Tidal will keep whatever value AI music adds while paying out nothing — and nobody has explained how the detector tells a Suno song from a sample-heavy human one.

What Tidal's AI policy actually commits to

Tidal — the streaming service that marketed itself on high-fidelity audio and paying artists better than Spotify — published an AI policy that tries to thread a needle: accept AI-generated music, but starve it. A reader pasted the email announcement verbatim, and the four commitments are concrete:

"Tidal will identify and tag AI-generated music … Listeners will see an 'AI' badge next to music we detect as wholly AI-generated … Tidal will not tolerate AI-generated music that impersonates an artist … No royalties will go to such releases, nor will AI-generated uploads be eligible for direct-to-fan sales." — a FOSS maintainer who quoted the announcement email in full

The policy defines its target as music "wholly or substantially generated by generative artificial intelligence" — and that word substantially is where the trouble starts. The grievance driving the whole thing is impersonation, not aesthetics. A subscriber spelled out the actual problem on the platform today:

"My Tidal 'feed' is full of new releases that are clearly AI-generated. They use the same artist name as artists that I really like, but the music is clearly not from the artist as advertised … I just don't want someone trying to spoof the artists I am interested in." — a Tidal subscriber describing his own feed

The two holes everyone poked: the definition and the detector

Approval was real — several readers want Spotify and YouTube to copy the demonetization rule wholesale. But two objections kept recurring, and both are practical, not ideological.

The first: nobody can say where "substantially" begins. A musician who'd made a living at it laid out the gradient the policy has to survive contact with:

"I know professional musicians that will use AI models like Suno as an aid … where the track may be x% AI and (1-x)% Human performance, where x is very small … But I also know people that use tools like Suno for everything … They basically just prompt some style and vibe they want." — a hobbyist musician who once made a living from it

The second hole is enforcement. If the badge and the royalty ban both hinge on detection, the policy is only as good as a classifier nobody has seen work.

"The detection problem is genuinely hard … detecting AI-generated music at the audio layer seems like a cat and mouse game that Tidal will struggle to win without self-reporting from uploaders." — a developer building desktop AI agents

That detector risks firing on the wrong people. EDM and hip-hop lean on sample libraries that are themselves quietly filling with AI-generated material, so an artist can be flagged for samples they didn't know were synthetic. And there's a cynical reading of the money itself: the U.S. Copyright Office has held that purely AI-made works aren't copyrightable, which gives Tidal a tidy argument for paying nothing — while still serving the tracks to subscribers whose fees it keeps. "Tidal has declared that AI music will now be more profitable to them," one reader noted dryly. The policy's fate rests entirely on a detector its critics doubt exists.

Get breakdowns like this every day

The top Hacker News stories, unpacked — by email, Telegram, or RSS.

Next story · 12 of 28 Read more
🔥Hot Discussion

An article calls the radiation safety model a lie. The thread fact-checks it line by line

Works in Progress argues a year of small radiation doses is far safer than the same dose all at once, and that regulators pretend otherwise. A geneticist who has sequenced Chernobyl thyroid tumors replied: show me the data.

★ 111 · 💬 86 · worksinprogress.co · 4 min