Spotify will tag AI-generated artist profiles with an AI Persona badge and, by default, keep those artists out of its recommendation systems, the company said Tuesday. The badge marks the identity behind a profile, not the song. An act that never existed as a person can still release music that stays on the platform, reachable by anyone who searches for it directly. The mechanism that actually decides an AI Persona’s reach is not the label. It is exclusion from recommendations.
TechCrunch reported Tuesday that artists can self-disclose starting August 11, 2026 through Spotify’s own artist dashboard, with the badges themselves appearing on profiles in mid-September. Once applied, the tag shows up in the profile banner, the About section, search results, and track rows inside playlists, so a listener encounters it wherever the artist appears rather than in one buried settings page.
Spotify says it will not lean on artists volunteering the label. It plans to review profiles directly, checking whether an artist’s name and photorealistic imagery represent a real person. Review begins with accounts that have already crossed unspecified audience thresholds, so the most-listened-to profiles get checked first. A public reporting tool for flagging unlabeled AI Personas is coming in the months after launch, and artists who believe they were mislabeled can appeal.
The recommendation cutoff has one exception: a user who actively follows an AI Persona will still see that artist’s music in their personal feed. Everyone else, meaning the algorithmic and editorial surfaces that introduce listeners to artists they did not search for by name, will not encounter that music unless they go looking. Spotify frames the badge as a statement about who is behind the microphone, not a verdict on how the track was produced. The company said the badge is “about the artist’s public identity, not about how the music was made.”
That split matters more than the wording suggests. Spotify does not have a single central chart; discovery on the platform runs almost entirely through algorithmic surfaces such as autoplay, radio, and personalized playlists that a listener never has to build by hand. An artist absent from all of them is not banned, technically. But for anyone who has not already typed that artist’s name into the search bar, absence from recommendations functions the same as absence from the catalog.
This policy extends one Spotify introduced in September 2025, which labels AI-generated tracks using shared industry detection standards and bars unauthorized AI voice clones. That earlier policy addressed how a song was made. This one addresses who appears to have made it, a narrower and arguably easier problem. Matching a profile photo and stage name against known-human patterns requires less certainty than judging whether a vocal chain used a cloned voice.
The distinction also has to hold up against a separate move: licensing deals with UMG and Merlin, the agency representing independent labels, that will let fans release AI-assisted remixes and covers with royalties flowing back to the original artists. Spotify needs listeners to tell an unlabeled AI Persona apart from an authorized AI remix of a human artist’s song, two categories that will otherwise look identical in a track row.
The timing lines up with a separate announcement the same day: Anthropic said it will begin watermarking text its models generate. Neither company is trying to stop machine-made content from existing. Both are building the plumbing to mark it once it is out, then let downstream systems, a chart, a feed, a search index, decide how much weight to give something once it is tagged as synthetic. For labels and distributors, the operating question for the next ninety days is not whether Spotify will delete AI act catalogs. It is how far down the discovery ranking those catalogs fall once the badges start appearing in September.
Reporting by Sarah Perez for TechCrunch, published August 11, 2026.
