Scale estimates by source
Fraud prevalence depends on definitions (what counts as "artificial"), where you measure (whole catalog versus charts), and whether you are looking at detected/confirmed fraud or estimated true fraud.
Platform disclosures
| Platform | Reported rate | Date | Notes |
|---|---|---|---|
| Spotify | <1% | Oct 2023 | Post-detection/confirmation claim |
| Deezer | ~7% | Sep 2023 | ML detection at user level |
| Deezer (AI tracks) | Up to 70% | Sep 2025 | Fraud rate on fully AI-generated tracks only |
Spotify told AFP that "less than 1% of all streams on Spotify have been determined to be artificial" and said manipulated figures are mitigated prior to payouts.
Deezer disclosed in its UMG partnership announcement that its algorithm identified approximately 7% of streams as fraudulent in 2022, using machine learning to detect fake accounts, payment fraud, and system-gaming behaviors.
In September 2025, Deezer reported that fully AI-generated music accounts for about 0.5% of total streams, but up to 70% of streams on those AI tracks are fraudulent.
France CNM study (2021 data)
The Centre national de la musique study provides the most rigorous government-backed analysis, using data from Spotify, Deezer, and Qobuz:
| Platform | Detected fraud rate |
|---|---|
| Spotify | 1.1% |
| Deezer | 2.6% |
| Qobuz | 1.6% |
The study found between 1 billion and 3 billion fake streams in France in 2021, representing 1-3% of total plays. Fraud was concentrated in the long tail (below top 10,000 tracks) on both Spotify and Deezer.
Warning The CNM noted that "the reality of fake streams exceeds what is detected." These are floor estimates, not ceiling estimates.
Vendor and industry estimates
Anti-fraud vendors like Beatdapp cite higher figures: at least 10% fraudulent streams globally and approximately $2 billion in annual losses.
Bias flag: Vendors have incentives to cite higher prevalence to justify anti-fraud spending. Treat these as upper-bound planning scenarios, not confirmed measurements.
Financial impact
Using the IFPI's reported $20.4 billion global streaming market value, fraud exposure scales by assumption:
| Fraud rate | Annual value at risk |
|---|---|
| 1% | ~$204 million |
| 3% | ~$612 million |
| 7% | ~$1.43 billion |
| 10% | ~$2.04 billion |
These represent value at risk under pro-rata royalty pools, not confirmed losses. Major platforms claim they withhold or mitigate detected fraud before payouts.
Royalty pool dilution
Most major DSPs operate on pro-rata pools: your share of total streams determines your share of the payout pool. Spotify notes that artificial streams, if undetected, dilute the royalty pool and shift revenue from legitimate artists to bad actors.
At 7% fraud (Deezer's disclosed rate), legitimate rightsholders lose approximately 7.5% relative to a fraud-free baseline. Dynamoi's Spotify royalty rate data shows what legitimate per-stream payouts look like when fraud is not diluting the pool.
Types of fraud
Artificial streaming (bots and automation)
Automated playback via bots, scripts, fake accounts, device farms, or credential-stuffed real accounts. This is the largest category by volume.
The CNM study found detected fake streams concentrated outside top charts (the long tail), consistent with many small-scale manipulations rather than blockbuster chart fraud.
Click farms and incentivized listening
Low-paid workers repeatedly playing tracks, sometimes blended with automation. Platform rules treat incentivized traffic as invalid.
Playlist manipulation
Selling playlist adds, inflating playlist follower counts, or operating playlists that are artificially streamed. This often overlaps with bot streaming because the playlist becomes the delivery vehicle.
Spotify explicitly states that paid services promising playlist placement for money violate terms and can lead to removals.
Pay-per-stream services
Vendors promising "X streams" or "guaranteed growth," usually powered by bots or compromised accounts. Both Spotify and Apple Music warn that anyone claiming they can deliver plays for a fee is unauthorized and typically uses fake accounts.
Content theft and impersonation
Reuploading copyrighted material or delivering tracks to another artist's profile to siphon royalties. Spotify describes impersonation attacks where uploaders fraudulently deliver music to established artist profiles.
Functional noise and spam flooding
Mass uploads of low-value audio (noise, ultra-short tracks, duplicates, SEO-name hacks) designed to extract payouts. Deezer reported "noise" content accounted for 2% of streams.
Spotify announced in September 2025 that it removed over 75 million spammy tracks in the prior 12 months amid the generative AI boom.
Platform detection and enforcement
Spotify
- Daily cleaning removes artificial streams from public counts
- Royalties withheld and numbers corrected when manipulation is confirmed
- Artificial streams do not count toward charts or boost recommendations
- Monthly reports to distributors about confirmed artificial streaming
- Per-track charges to distributors for "flagrant" artificial streaming
Spotify's September 2025 announcement introduced a new spam filter that identifies uploaders and tracks engaging in manipulation tactics, tags them, and stops recommending them.
Deezer
- 7% of streams identified as fraudulent in 2022 via user-level ML
- Up to 70% fraud rate on fully AI-generated tracks (2025)
- Manipulated streams excluded from royalty payments
- AI-tagging initiative to identify and label generated content
Apple Music
Apple Music warns of an industry-wide increase in artificial streaming tactics but provides less detailed public disclosure than Spotify or Deezer. Reporting suggests Apple claims less than 1% fraudulent streams, though Apple's own documentation is thinner on specific numbers.
YouTube
YouTube addresses the problem as "invalid traffic":
- Fraudulent and incentivized boosting classified as invalid
- Advertisers refunded when invalid traffic detected
- Creator earnings withheld, adjusted, or offset
- Channels may be demonetized; accounts can be disabled
Legal enforcement (documented cases)
Legal actions provide concrete "how big can it get" anchors:
United States (2024)
Federal prosecutors charged a North Carolina musician with fraudulently obtaining over $10 million in royalties through AI-generated music and bot streams. At its height, the scheme allegedly generated approximately 661,440 streams per day across Amazon Music, Apple Music, Spotify, and YouTube Music. Notably, Spotify stated its preventive measures limited the royalties from this scheme to approximately $60,000 of the $10 million total.
Denmark (2024)
A Danish man was convicted of fraudulently earning at least 2 million DKK (approximately $290,000) from artificially generated streams of hundreds of tracks. The case also involved altered works published under false authorship.
Brazil (2025)
IFPI-supported operations in Brazil have resulted in multiple enforcement actions:
- Operation "Out of Tune" (March 2025): Individual arrested for 28 million fake plays across 400 fake tracks, generating over $65,000 with $400,000 in assets seized
- JustAnotherPanel disruption (April 2025): Court order blocked a global fake-stream service, disrupting 43 local services and impacting 1,131 foreign resellers
- Operation Authentica (July-October 2025): Civil court rulings against companies offering fake followers and streams across Spotify, YouTube, TikTok, and Instagram
Red flags for artist teams
Streaming pattern warnings
Spotify identifies these as abnormal activity signals:
- Sudden spike in streams "for no apparent reason," followed by a drop-off
- Geographic spike from a location where you have never had traction
- Most streams from odd sources (e.g., "Other" or unexpected playlists)
The CNM study describes platform detection logic focusing on abnormally strong periods, very long continuous listening sessions, and unusually high stream volume patterns.
Audience integrity signals
- Short-lived follower spikes that do not match saves, shares, or social growth
- Public versus private metric discrepancies (Spotify for Artists data may temporarily show spikes that public counts later correct)
Promotion service red flags
Avoid anyone guaranteeing plays or playlist placement for a fee. Both Spotify and Apple explicitly warn against this.
Due diligence checklist:
- Require services to explain exactly what they do (press outreach, influencer seeding, paid social, playlist pitching with no guarantees)
- If they cannot describe methods without "secret sauce," treat as high risk
- Prefer campaigns you can audit: paid social receipts, creator lists, PR coverage, UTM links
- Watch for services pushing volume-first KPIs ("50k streams") instead of fan-intent KPIs (saves, follows, repeat listeners)
Trends and outlook
AI has shifted fraud tactics
Fraud has moved from "repeat one track" to "flood the catalog." Deezer reports rapid rise in fully AI-generated uploads with high fraud rates. Spotify describes AI-enabled content farms and spam tactics, and has removed 75 million+ tracks in response.
Detection is increasingly about upload behavior and network signals, not only suspicious play loops.
Platforms are escalating deterrents
Financial penalties (per-track charges for flagrant fraud), algorithmic penalties (removed from recommendations), and content removal are all increasing.
Fraud tactics are evolving
Reporting notes more sophisticated abuse including using credentials from legitimate user accounts to direct streams, complicating detection versus obvious device farms.
Real-world enforcement is accelerating
The 2024-2025 period includes multiple law enforcement actions: US federal charges, Danish conviction, and IFPI-backed operations in Brazil.
Key benchmarks for planning
| Metric | Estimate | Source | Confidence |
|---|---|---|---|
| Spotify detected fraud | <1% | Spotify (Oct 2023) | Platform claim |
| Deezer detected fraud | ~7% | Deezer (Sep 2023) | Platform disclosure |
| France detected fraud | 1-3% | CNM (2021 data) | Government study |
| Vendor upper estimates | 10%+ | Beatdapp et al. | Vendor claim |
| AI track fraud rate | Up to 70% | Deezer (Sep 2025) | Platform disclosure |
| Spam tracks removed | 75M+ | Spotify (Sep 2025) | Platform disclosure |
| Global market at 1% fraud | ~$204M | Derived | Calculation |
| Global market at 7% fraud | ~$1.43B | Derived | Calculation |
The bottom line: fraud exists at material scale (somewhere between 1% and 10%), detection is improving but not complete, and the safest posture for artist teams is to avoid any promotion service that guarantees streams or playlist placement for a fee.