AI music just crossed a new line: Deezer says nearly all listeners can't tell machine-made songs from human ones, even as the platform drowns in AI uploads and fraud.
In a new Ipsos survey of 9,000 people across eight countries, 97% of respondents failed to reliably distinguish AI-generated songs from human tracks; at the same time, Deezer reports more than 50,000 fully AI tracks hitting its system every day, roughly a third of all new deliveries.
Deezer now excludes 100% AI tracks from editorial playlists and algorithmic recommendations and says up to 70% of plays on those tracks are fraudulent streams it scrubs from royalty payouts.
Why it matters:
For labels, distributors and marketing teams, this isn't an abstract AI debate; it's a direct hit to your release strategy, royalty economics and ad planning.
If listeners can't hear the difference but 40% say they'd skip AI tracks when given the choice, the real battle becomes transparency, trust and where platforms decide to point their recommendation firehose.
- Royalty pools are changing: When a DSP filters out fraudulent AI streams and quarantines fully synthetic tracks from recommendations, the effective revenue share for remaining catalog quietly shifts.
- Playlist strategy gets riskier: If your release schedule leans heavily on AI-assisted or synthetic projects, they may never touch editorial or algorithmic placements on Deezer-style platforms.
- Brand safety moves to music: For sync, influencer and UGC campaigns, brands will start asking whether the soundtrack is human, AI or hybrid — and whether that matters to their audience and legal team.
By the numbers:
- 50,000+ fully AI-generated tracks are delivered to Deezer every day, up from 30,000 in September and just 10,000 in January.
- 34% of all daily track deliveries to Deezer are now fully AI-generated.
- 97% of listeners in an eight-country survey could not reliably tell human and AI songs apart.
- 73% want clear labeling when AI tracks are recommended, 45% want filtering controls and 40% say they'd avoid AI music entirely if they can.
- 70% of plays on fully AI-generated tracks are flagged as fraud, and those streams are excluded from royalty payments, leaving AI tracks at just 0.5% of total streams despite their upload volume.
Zoom in:
Deezer isn't banning AI outright; it's drawing a bright policy line between 'fully AI' and everything else, then wiring that line directly into curation, recommendations and royalty logic.
What labels and distributors should do now
- Lock your definitions: Create a clear internal taxonomy for human, AI-assisted and fully AI-generated tracks, and align it with how each DSP is actually treating those categories.
- Audit your pipes: Check with your distributor which fields they send to Deezer (and peers) around AI usage, and whether any catalog is being silently classified as 'fully AI' without your team realizing.
- Model AI-off scenarios: Run revenue projections that assume fully AI tracks get zero algorithmic support and severe fraud filtering, then stress-test your 2026 release calendar against that reality.
How this hits marketing teams
If AI tracks are increasingly boxed out of recommendations and playlists, paid and organic campaigns that lean on them will behave very differently by platform.
- Reframe your A/B tests: Treat AI vs human vocals, production and songwriting as segmentation experiments, not just creative flair, and track downstream playlisting and royalty impact per version.
- Use 'human-made' as a positioning lever: With almost half of listeners wanting the ability to avoid AI music, campaigns that lean into human craft — session photos, behind-the-scenes clips, songwriter credits — can differentiate in a saturated feed.
- Instrument your fraud risk: Coordinate with your fraud-monitoring vendors and distributors to ensure your campaigns aren't accidentally juicing bots on tracks that will later have streams clawed back.
The bottom line:
The story here isn't that AI music sounds good or bad — it's that zero-marginal-cost content plus opaque streaming economics creates irresistible incentives for fraud.
Deezer's move turns that from a theoretical worry into concrete policy, and once one DSP bakes anti-fraud AI rules into playlists and payouts, others have political cover to follow.
For artist teams, that means two parallel roadmaps: embracing AI where it actually compounds human creativity, and building governance, metadata and fraud controls robust enough that your catalog doesn't get caught in the next wave of DSP crackdowns.