44% AI – Deezer says 44% of daily song uploads are AI-generated, but they account for just 1–3% of streams—while most streams are flagged as fraudulent.
Deezer’s AI-upload surge, but streaming remains limited
Deezer has put a spotlight on how quickly AI-generated music is moving from novelty to volume on streaming platforms, with AI tracks making up **44% of all new music uploads daily**.
Misryoum breaks down what those numbers mean for listeners, artists, and the economics of music streaming—especially as platforms tighten detection, labeling, and monetization rules.
Deezer estimates it receives **nearly 75,000 AI-generated tracks per day**, and **more than two million per month**.. Those uploads have risen fast since Deezer first introduced an AI-music detection tool. with daily AI-track deliveries climbing from roughly 10. 000 at launch to around 60. 000 by January.. The company’s figures suggest not just an uptick in experimentation by creators. but also a growing “supply pipeline” of AI music entering catalogues.
The core tension: high uploads, low listener pull
Despite the sharp rise in AI submissions, Deezer says consumption is still relatively small. AI-generated music accounts for just **1–3% of total streams** on the platform.
That gap—between uploads and actual listening—matters.. It points to a system where much of the content arriving may not be meeting listener demand. passing quality thresholds. or proving legitimate enough to be monetized.. In Deezer’s case, the company reports that **85% of AI-related streams are detected as fraudulent and demonetized**.
This is one of the most consequential details in Deezer’s update.. When a large share of activity connected to AI tracks is treated as fraud, it changes the business equation.. Streaming revenue becomes harder to predict. royalty flows are disrupted. and legitimate creators face higher scrutiny as platforms try to distinguish authentic output from content engineered to game distribution.
How Deezer reduces exposure to AI content
Deezer says AI-tagged tracks are automatically removed from algorithmic recommendations and are not included in editorial playlists. The platform is also tightening technical handling of this category of content: it announced it will **no longer store hi-res versions of AI tracks**.
For everyday listeners. those design choices likely translate into fewer AI tracks surfacing through the same discovery channels as human-made music.. For the industry. it signals that transparency and categorization are no longer just policy statements—they’re operational changes that can affect reach. audio quality availability. and ultimately the likelihood a track finds an audience.
The broader challenge is that AI music can look, sound, and even “feel” complete enough to pass casual listening tests.. Deezer’s own survey work last November found that **97% of participants said they couldn’t tell the difference** between fully AI-generated music and human-made music.. Another finding showed **80% want clear labeling**, while **52% believe 100% AI-generated songs shouldn’t appear in mainstream charts alongside human-made releases**.
Fraud detection and why it may shape future music economics
Misryoum interprets Deezer’s approach as a response to a growing risk: AI tools can lower the cost of producing tracks at scale. which can invite bad-faith behavior.. If malicious actors can generate large volumes quickly. they may attempt to inflate plays. dilute payouts. or otherwise disrupt royalty systems.
That’s why Deezer’s reported demonetization rate is central.. If most AI-related streams are being flagged as fraudulent. the implication is that the platform has to devote significant effort to verification and filtering—because simply hosting more content doesn’t equal healthier music economics.. It also means legitimate AI creators may find themselves inside a detection net designed for abuse.
There’s also an ecosystem-wide knock-on effect.. When platforms demote or restrict certain categories of tracks, the incentive structure changes for distributors and labels.. Even without inventing details. the direction is clear: the next phase of streaming competition is likely to be less about sheer catalogue size and more about trust—who gets paid. how plays are verified. and how clearly users understand what they’re hearing.
A labeling moment—because the public can’t always tell
Deezer’s update comes at a time when AI-generated tracks are making visible headlines. including a track that reportedly topped music charts across multiple countries.. Whether listeners identify those tracks as AI or not. the commercial reality is that AI music is now capable of reaching mainstream attention.
Misryoum sees the survey responses as a useful compass.. If most people can’t reliably distinguish AI from human music. then labeling becomes less of a “nice to have” and more of a fairness mechanism.. Without disclosure, charts and recommendations can become marketing channels where the audience doesn’t understand what’s driving the recommendation.
Deezer also said it first began tagging AI tracks at the platform level in June 2025, and it claims it tagged more than **13.4 million AI tracks** across 2025. The volume underscores how quickly AI content can accumulate once a platform detects and categorizes it at scale.
What other platforms—and the industry—may do next
Deezer’s announcement sits within a wider pattern of experimentation across streaming services.. Some platforms rely on distributor-side filtering and transparency efforts, while others move toward detection and tagging inside their own systems.. Misryoum expects the next competitive wave to center on three questions: how reliably AI content is identified. how consistently it is labeled. and how monetization rules apply when fraud is detected.
If AI-generated music continues to grow in uploads faster than it grows in legitimate listener streams. platforms may keep tightening restrictions—especially around recommendations and monetization—until verification improves.. At the same time. consumer demand for clear labeling could pressure platforms to treat transparency as a baseline feature rather than an optional setting.
For artists—AI-assisted or fully AI-generated—the practical stakes are high.. Discovery rules can determine whether a track gets heard or disappears into the margins.. For listeners. the key issue is confidence: knowing whether the music they’re streaming reflects the ecosystem they believe they’re supporting.
The headline number—**44% of daily uploads** being AI-generated—may be startling, but the story underneath is more nuanced: uploads are surging, mainstream listening remains limited, and fraud controls are becoming a defining part of streaming’s response.
United–American merger warning: Senators fear higher fares and weaker competition