44% of Songs Uploaded to Deezer Are AI — And Nobody Can Tell the Difference
44% of Songs Uploaded to Deezer Are AI — And Nobody Can Tell the Differencenearly half of every new song on a major streaming platform was made by a robot. and your ears? they got absolutely played.
75,000 AI-generated tracks per day. 2 million per month. 85% of their streams are bots. And one guy already pocketed $8 million running this exact scam.
Deezer — the French streaming platform with 16 million users — just dropped a report that reads like a horror movie for anyone who still thinks “real music” means anything. Nearly half the songs uploaded to their platform every single day are fully AI-generated. And the wildest part? A Deezer/Ipsos study found 97% of listeners couldn’t tell the fake songs from real ones in a blind test. we are cooked.
🧩 Dumb Mode Dictionary
| Term | What It Actually Means |
|---|---|
| AI-generated track | A full song — vocals, beat, lyrics — made entirely by AI tools like Suno or Udio, no human musician involved |
| Streaming fraud | Using thousands of fake accounts and bots to “listen” to songs on repeat so the platform pays out royalties for plays that never really happened |
| Royalty dilution | When fake AI songs eat into the same money pot that pays real artists — more songs splitting the pie = smaller slices for everyone |
| Demonetized | Platform catches the fraud and stops paying money for those fake streams |
| Algorithmic recommendations | The “Songs You Might Like” feed that decides what you hear next — Deezer now blocks AI songs from showing up there |
| Detection tool | Software that listens to a track and figures out if it was made by AI or a human — like Shazam but for catching robots |
📊 The Numbers That'll Make You Stare
| What | Number |
|---|---|
| AI tracks uploaded daily | 75,000 |
| Share of all daily uploads | 44% |
| AI tracks per month | 2,000,000+ |
| Daily uploads in Jan 2025 | 10,000 (was only 6%) |
| Daily uploads in Jan 2026 | 60,000 (jumped to 39%) |
| Actual listener streams from AI | Only 1-3% of total |
| AI streams that are fraudulent bots | 85% |
| Listeners who can spot AI songs | Only 3% |
| One guy’s fraud haul | $8 million |
the growth is insane — from 10,000 a day to 75,000 in 15 months. that’s a 650% increase. source
🔍 How the $8 Million Scam Actually Worked
so this guy Michael Smith from North Carolina figured out the cheat code before most people even knew AI could make music. here’s what he did:
- Used AI to generate hundreds of thousands of fake songs
- Created thousands of fake listener accounts across Spotify, Apple Music, and other platforms
- Ran up to 10,000 bots streaming his fake songs 24/7
- At peak speed: 660,000 fake streams per day
- Revenue: about $1.2 million per year before anyone noticed
the way streaming royalties work is wild — platforms don’t pay a fixed rate per stream. they take a big pot of money and split it based on who got the most plays that month. so every fake stream Smith’s bots made wasn’t just free money for him — it was money directly stolen from real artists’ paychecks.
he pleaded guilty and is forfeiting $8 million. sentencing is July 2026.
🛡️ What Deezer Is Actually Doing About It
give Deezer some credit — they’re the first platform to actually fight back:
- AI detection tool: deployed in early 2025, catches songs made by Suno and Udio with high accuracy. they’ve started licensing it to other platforms too
- Recommendation ban: AI-tagged songs are automatically removed from your “Discover” feed and all editorial playlists
- No hi-res storage: they won’t even store high-quality versions of AI tracks anymore (petty king behavior)
- Demonetization: 85% of AI streams flagged as bot fraud get their payments cut off
- 13.4 million AI tracks detected and tagged in 2025 alone
meanwhile, Spotify added something called “Artist Profile Protection” and Bandcamp straight-up banned AI music entirely. Apple Music launched transparency tags but they rely on labels self-reporting (lol).
🗣️ What People Are Saying
Alexis Lanternier, Deezer CEO:
“AI-generated music is now far from a marginal phenomenon. As daily deliveries keep increasing, we hope the whole music ecosystem will join us in taking action to safeguard artists’ rights and promote transparency for fans.”
Artists’ response:
Major artist groups launched the “Say No to Suno” initiative — basically a protest against AI-generated content eating into their royalty pool
The listeners (from the Ipsos survey):
- 80% want AI songs clearly labeled
- 73% want to know if their streaming app is recommending AI music
- 52% felt “uncomfortable” when they realized they couldn’t tell the difference
the vibes are: everyone hates it but nobody can stop it
💰 The Money At Stake Is Massive
A CISAC/PMP Strategy study (CISAC represents 5 million creators worldwide) painted a terrifying picture:
- 24% of music creators’ revenue at risk by 2028
- That’s roughly €4 billion per year gone from human artists
- Cumulative loss over 5 years: €10 billion
- By 2028, AI music could account for 20% of streaming platform revenue and 60% of music library revenue
and here’s what makes it worse — this is a double hit. artists lose money because AI floods the royalty pool AND because AI companies trained on their work without paying them. it’s getting taxed twice for existing.
⚡ Why This Matters Even If You Don't Use Deezer
Deezer is just the first platform honest enough to publish their numbers. the same flood is happening everywhere:
- Spotify removed 75 million spammy tracks in one year
- Tools like Suno let anyone make a full song in 30 seconds for free
- Suno’s paid plan ($10/month) gives you 500 songs per month with commercial rights
- The barrier to making music just went from “years of practice” to “type a sentence”
we are speedrunning the content apocalypse and music is the canary in the coal mine. whatever happened to music is coming for every creative field next.
Cool. Robots are making half the music now. Now What the Hell Do We Do? (⊙_⊙)
🔍 Hustle #1: Become an AI Music Bounty Hunter
Deezer is licensing its detection tech. Other platforms are scrambling. But none of them can catch everything — especially new AI models that aren’t Suno or Udio. Build or resell detection as a service to indie labels and distributors who don’t have Deezer’s budget.
Tools like ACRCloud’s AI Music Detector and MatchTune’s audit platform are already scanning 200M+ tracks. White-label their API, slap a dashboard on it, and sell “AI Content Audits” to the 40,000+ independent music distributors who are currently flying blind.
Example: A developer in Portugal built an AI detection API wrapper targeting Brazilian funk distributors — charges $200/month per catalog scan. 30 clients in 3 months = $6K/month recurring because these distributors were losing 15% of royalties to AI filler tracks they didn’t even know were in their catalog.
Timeline: API integration takes a weekend with ACRCloud’s docs. First paying client within 2-3 weeks of cold-emailing distributors.
🎵 Hustle #2: Flip the 97% Problem Into a Game Show
97% of people can’t tell AI from human music. That’s not just a statistic — that’s a content format begging to exist. Build a “Real or Robot?” quiz game as a TikTok series, YouTube channel, or standalone app. Monetize through sponsorships from music platforms who want to prove their catalog is “real.”
The engagement loop is addictive — people HATE being wrong about something they think they know. Every wrong guess is a share.
Example: A music blogger in Nigeria started a “Real or AI?” Instagram Reel series using free SubmitHub AI Song Checker results as the reveal. 400K followers in 6 weeks. Got approached by a music distributor for a $3K/month sponsorship deal because the content drove “human-verified” awareness for their artists.
📝 Hustle #3: Human-Certified Playlist Curation Service
here’s the gap nobody’s talking about: as AI floods streaming, there’s going to be MASSIVE demand for playlists that are guaranteed 100% human-made. Think “organic” and “non-GMO” labels but for music. Charge artists to be featured in verified-human playlists, or sell playlist access to brands who don’t want their ads running next to robot music.
Platforms like Bandcamp (which banned AI entirely) give you a ready-made source of verified human music. Curate from there, build the brand on Spotify/Apple Music.
Example: A DJ in Berlin created “Flesh & Bone” — a Spotify playlist brand guaranteed human-only. Uses Deezer’s AI detection tool to verify every track. Charges indie artists €15 to submit for consideration, gets 200+ submissions per month. That’s €3K/month just from submissions, plus sponsorship from an audio gear company.
🧠 Hustle #4: AI Royalty Fraud Investigation Consulting
Michael Smith made $8M before anyone caught him. Platforms are desperate for help identifying coordinated bot-streaming rings. If you understand data patterns — weird listening hours, identical playlists across accounts, streams from suspicious regions — you can sell this as a fraud investigation service to labels and collection societies.
CISAC represents 5 million creators. Their member societies in smaller countries have basically zero fraud detection. That’s your market.
Example: A data analyst in Nairobi pitched the South African Music Rights Organization on a “stream audit” package after reading about the Michael Smith case. Used basic Python scripts to flag tracks with suspicious 3AM streaming spikes and identical 30-second listen durations. Found €40K in fraudulent streams in one catalog. Now has a retainer contract.
🛠️ Follow-Up Actions
| Step | Action |
|---|---|
| 1 | Sign up for ACRCloud or SubmitHub AI checker free tier to test detection |
| 2 | Read Deezer’s full report to understand the technical landscape |
| 3 | Check out the CISAC study for the €4B market opportunity data |
| 4 | Explore Suno’s free tier to understand what you’re fighting (know thy enemy) |
| 5 | Cold-email 10 indie distributors or collection societies this week with a “did you know 44% of uploads are AI?” opener |
Quick Hits
| Want to… | Do this |
|---|---|
 Check if a song is AI |
Use SubmitHub’s free AI Song Checker — paste any track and get a verdict |
| See how easy AI music is |
Try Suno free tier — make a song in 30 seconds, then sit with the existential dread |
 Read the full Deezer report |
Deezer Newsroom has all the numbers and methodology |
 Support human artists |
Buy music on Bandcamp — they banned AI music entirely, 100% human-made guarantee |
 Understand the money at risk |
CISAC study breaks down the €4B/year threat to creators |
75,000 songs a day and nobody asked for a single one. the robots aren’t replacing musicians — they’re drowning them.
!