30 Jun Tidal no longer pays for AI music – Deezer counts 44 percent AI uploads
▶ 6:20 reading time
Tidal has made its decision: Fully AI-generated tracks will no longer earn royalties. The music isn’t banned-it just won’t make any money. This might sound like streaming bureaucracy, but it’s the first hard line against an upload flood that almost nobody listens to. Deezer reports 44 percent of uploads are AI-generated-nearly 75,000 tracks a day. The few streams these tracks get are largely fake. In this gap between volume and demand, Tidal, Deezer, Spotify, Apple Music, and Bandcamp are defining how they handle AI.
How Tidal is removing AI tracks from the royalty pool
The new policy went live on June 29. Fully AI-generated tracks no longer earn royalties and are also excluded from direct sales to fans. The music remains on the platform-it just doesn’t generate revenue. Starting July 15, every track Tidal identifies as purely AI-generated will carry a visible AI badge. Later, the label will expand to include predominantly AI-generated music.
When it comes to fraud, Tidal is drawing a harder line. From mid-July, the service will block or remove AI music linked to fraudulent activity. This includes tracks that deceive listeners, imitate real artists’ voices or names, or are boosted through mass uploads and suspicious streaming patterns. Tidal is also shifting responsibility to distribution partners: they must label AI content before it even reaches the platform.
44 Percent Uploads, Almost No Listeners
If you want to understand why platforms are finally taking action, look at Deezer. The French service released figures in April that put the debate into measurable terms. Nearly 75,000 AI-generated tracks are uploaded daily-around 44 percent of all new uploads. That’s over two million AI tracks per month.
The more critical part often gets overlooked: almost none of it gets played. Only 1 to 3 percent of streams come from AI music, because Deezer removes identified tracks from recommendations. Of those few streams, around 85 percent are flagged as fraudulent and demonetized. The issue isn’t listeners suddenly embracing AI music. The problem is thousands of uploaded tracks siphoning money from the royalty pool meant for real artists through fake streams.
This is the key to everything Tidal, Spotify, and others are doing right now. Hardly any platform is fighting AI as a tool-they’re fighting mass uploads, stream farming, and identity theft. This distinction is practical: it determines who gets hit by the measures.
The AI takeover of the music industry isn’t inevitable-if we act decisively now to monitor and control it.
Tony Gervino, Tidal
Gervino sounds combative. In reality, he’s describing an economic decision: Tidal is protecting the royalty pool from automated dilution. That’s reasonable. But it’s not proof that streaming services are suddenly saving culture. Keep this distinction in mind the next time a platform sells its AI policy as a commitment.
How Five Platforms Are Drawing the Line
Bandcamp is taking the hardest stance. Since January 13, music that is entirely or largely AI-generated has been banned. It’s the only outright prohibition among the major players. But Bandcamp isn’t a discovery service with recommendation algorithms-it’s a marketplace where fans buy directly from artists. A pure stance like this is easier to enforce than it would be for a streaming giant with hundreds of millions of tracks.
Deezer is the tech pioneer. The platform was the first to visibly tag AI-generated music in June, flagging over 13 million tracks last year. Its in-house detection tool identifies output from Suno and Udio. Deezer now licenses this technology to others. Here, the approach has real substance: detection plus demonetization of fraud delivers measurable results.
Spotify is cleaning house. Over 75 million spam tracks removed in twelve months, a spam filter against mass uploads, protection against unauthorized voice clones, and since April, a feature allowing artists to self-declare AI use in song credits. No outright ban, but clear pressure on farm accounts.
Tidal is taking the middle ground you saw above. Allowed, but not paid. Plus a badge, plus removal for fraud and imitation.
Apple Music is Bandcamp’s opposite-from the other side. Its transparency tags, introduced in March, sound progressive but are voluntary. Labels and distributors can declare AI use in artwork, tracks, compositions, or music videos, but they don’t have to. If the tag is missing, nothing is assumed. A labeling system no one is required to use ultimately only marks the honest.
What is fully AI-generated music? It refers to music where composition, instrumentation, and vocals come entirely from models like Suno or Udio, with no human recording involved. The line blurs when you work hybrid-layering an AI voice over your own beat, for example, or separating stems with AI. This gray area is why Tidal plans to apply its badge first to fully AI-generated tracks, then later to predominantly AI-generated ones. Where that line is drawn is a political question, not just a technical one.
What You Must Disclose When Using AI in Your Workflow
Let’s get specific-most producers aren’t running server farms. You might use an AI tool to separate stems, get a bassline idea suggested, or test out a generated vocal track. That’s perfectly fine. None of the five major platforms ban this. The real risk is different: technically, you’re lumped in with mass uploads if your metadata is sloppy or the labeling is missing.
Three key points emerge from this. First, it’s better to separate the question of *allowed* from *paid*. On Tidal, you can upload AI-assisted tracks, but a fully generated one won’t earn you royalties or direct sales. Don’t plan your income around pure AI production. Second, labeling becomes mandatory. Deezer’s tags, Spotify’s song credits, Tidal’s badges, and the DDEX standard all point in the same direction. If you work hybrid, clarify what your distributor and platform require *before* release. Undeclared AI isn’t a minor oversight anymore-it’s a fraud risk. Third, clean metadata and a genuine artist identity become a visibility issue. No misleading credits, no borrowed artist names, no disguises. This is the line algorithms use to decide whether you appear in recommendation feeds or vanish into the spam bucket.
- you use AI as a tool while producing human-made content
- your metadata and credits are clean
- you transparently declare AI contributions
- your income model relies on pure AI production
- you skip labeling because no one checks
- you use borrowed voices or names
Post-Show Q&A
Click on a question to reveal the answer.
Is AI-generated music being banned now?
Will listeners even notice any of this?
I use AI tools in my production. Does this affect me?
What happens to hybrid tracks after July 15?
AI builds the 90-percent beat. What’s left for you? →Stem separation: What AI tools can really do →
Image source: AI-generated (June 2026)
Also available in