Spotify's Plan to Push AI Personas Out of Automated Recommendations
Spotify will start removing AI-generated artist identities, which it calls 'AI personas', from automated recommendation surfaces from mid-September. That includes algorithmic playlists such as Discover Weekly and Release Radar, as well as Spotify radio streams. The music will remain on the platform, but listeners will only encounter it if they explicitly follow an AI profile or search for it directly.
The change responds to a surge of machine-made music. Competing streaming service Deezer has said that half of the tracks uploaded daily in June, about 90,000 songs, were AI-generated. Spotify also plans new in-app labels so users can see in search results and on title pages when a profile is classified as an AI Persona, something previously limited to blue verification badges for major artists.
Spotify says it will identify these profiles through voluntary declarations from producers, its own review teams and automated analysis. A user reporting function is expected in the coming months. The company stressed that a human artist who uses AI as a creative tool will not automatically be reclassified as an AI Persona.
Music creators' groups welcomed the step. But unresolved issues remain, especially bot farms that inflate streams and the industry's lack of uniform rules for when AI-generated or AI-assisted music must be labeled.
How Spotify's Algorithm Change Shifts Royalties, Labels and Discovery
Where Spotify Is Applying Its Real Leverage
The move does not ban AI music, but it cuts off the main route to accidental discovery. Previously, AI background tracks reached millions through algorithmic playlists without listeners realizing no musician was behind them. By separating AI personas from recommendation surfaces, Spotify is using the same mechanism that built modern streaming hits to contain synthetic output.
The Royalty Shift That Could Help Small Human Artists
Streaming royalties are paid pro-rata from total streams. Marketing professor Dominik Papies told Tagesschau that AI acts could face an income problem once they lose playlist placements and automatic streams. He said small human artists in particular may benefit, potentially financially, if AI tracks stop absorbing streams. Still, researchers at the University of Chicago found 93 percent of AI music is barely or never played, which limits how large that shift can be.
The Definitional Gray Area: AI-Assisted vs. AI-Generated
Spotify's rule targets artist profiles, not creative tools. A human artist using AI to make a track will not automatically become an AI Persona. But industry labels create new disputes: under BVMI guidance, a work is considered AI-generated if vocals or key instrumental parts were prompted, while AI-written lyrics sung by a human do not require labeling. That distinction has already drawn criticism from musicians and lyricists.
What the Policy Does Not Solve
Bot farms, computers that play songs on loop to generate streams, remain outside this change. Papies and researchers at the Vienna University of Economics and Business are studying how many streams are bot-driven, and the practice continues to harm labels and artists. There are also no uniform cross-platform rules; the voluntary codes agreed by music associations in July are only a starting point.
What Artists, Labels and Listeners Should Expect From September
For artists and labels:
- Human artists using AI tools should document how AI was used, because Spotify's classification will rely on self-disclosure, review teams and automated checks, and the company has said AI used as a creative tool will not alone trigger AI-Persona status.
- Labels and distributors should clarify whether catalog entries fall under the industry's new AI-generated or AI-assisted labels, since voluntary self-disclosure is one of the first filters Spotify will use.
- Prepare for the promised user reporting function later this year, and decide who inside your team will respond if a human profile is mistakenly flagged as an AI Persona.
For listeners:
- From mid-September, expect fewer AI-generated personas in Discover Weekly, Release Radar and Spotify radio; look for the new label in search results and on title pages to check whether a profile is an AI Persona.
Risk & Opportunity Assessment
| Commercial Risk | Medium | AI persona profiles lose algorithmic reach and listening-driven royalties, but the impact may be limited because content remains searchable and 93 percent of such music is barely played. |
| Competitive Risk | Medium | Spotify's stricter labeling could differentiate it from streaming rivals with less transparent rules, but it may also push AI uploaders toward platforms that still give algorithmic reach. |
| Regulatory Risk | Low | The policy builds on voluntary industry codes rather than government enforcement, though future EU or national AI rules could impose stricter requirements. |
| Reputation Risk | Low | Transparency is likely to be viewed positively, but disputes over AI-written lyrics not requiring labels may fuel criticism from musicians and lyricists. |
| Technology Disruption | High | Deezer reports that half of daily uploads in June, around 90,000 tracks, were AI-generated, showing AI production has become a mainstream publishing force. |
| Commercial Opportunity | High | Removing AI personas from recommendations may return exposure and pro-rata royalties to smaller human artists, as Dominik Papies argued. |
Comments 0