The Hook
By late June 2025, a band called The Velvet Sundown had two albums on Spotify and approximately 1.4 million monthly listeners. The music was mellow, competent indie rock — the kind that populates Spotify's algorithmic playlists reliably. Then Reddit users noticed that none of the band members were findable outside the Spotify profile. The project subsequently admitted the music was generated by Suno.
The Question
When a recommendation algorithm surfaces AI-generated music to millions of users without disclosure, which failure happened — the generation or the distribution?
The Paper Trail
Spotify's Discover Weekly algorithm surfaces tracks based on listening-pattern matching, not origin verification. At the time The Velvet Sundown was circulating, Spotify had no mandatory disclosure requirement for AI-generated audio. The discovery was made by users on Reddit, not by the platform.
On September 25, 2025, Spotify changed its terms: AI-generated content requires disclosure, and voice cloning of identifiable artists is prohibited. The platform also reported removing approximately 75 million tracks flagged as AI-generated spam in a twelve-month period. Both actions confirm the platform was aware of the scale of the problem before the policy change.
The Synthesis
The Velvet Sundown story is not primarily about the music. It is about the recommendation layer: the same algorithm that surfaces a genuinely unknown band to millions of listeners will surface a generated band to the same listeners with the same signal. Authenticity is not a variable in the engagement optimization.
The policy change Spotify made — disclosure requirement, spam removal — demonstrates that the platform had the ability to act before the Reddit users found the gap. The decision to wait is a product choice, not a technical limit.