On Deezer, roughly half of everything uploaded on a given day is now machine-made. That statistic is the backdrop to Spotify’s latest announcement, and it explains why the company has taken a step it plainly would rather have avoided: tagging the artists themselves.
From mid-September, any artist profile that does “not represent a real person” will display an “AI Persona” badge throughout the app. By default, tracks from those profiles will be kept out of both editorial suggestions and personalized recommendations.
They aren’t being demoted or pushed down the rankings. They’re being cut from the surfaces where the majority of listening sessions begin.
Spotify isn’t waiting for artists to raise their hands
As of today, artists can voluntarily declare themselves an AI persona. The flaw in any self-identification scheme is that whoever is running a slop farm has zero reason to come forward, and Spotify appears well aware of that.
Hence the company’s plan to combine human review with AI tooling to determine whether a profile — its name and image included — “appears to represent photorealistic AI-generated identities.” Where it concludes an identity is probably AI-generated but was never disclosed, a gentler tag goes on instead: “Likely AI Persona.”
That hedge is worth noting. One badge records something the artist confirmed. The other is Spotify’s estimate — and Spotify is openly flagging it as an estimate.
The big accounts get checked first
Priority in the review queue goes to artists who meet “defined audience thresholds,” Spotify says, so that the most-streamed profiles are covered first. It’s reasonable triage, and simultaneously a concession about volume. No one is reviewing the full catalog by hand.
The language from the company is unusually direct. “Listeners have been clear in telling us that they don’t like seeing an artist profile that seems human, only to find out that the persona is AI-generated,” Spotify says. “We tune our algorithms to highlight authentic human artists and, based on available signals, strive to not include spam, slop, or low-effort content in our recommendations.”
Spam. Slop. Low-effort content. That’s a streaming platform characterizing a sizeable share of what gets uploaded to it.
The label is about who, not how
This is the boundary to keep in mind: the new labels only pertain to an artist’s identity, not how they created their music. Push an entire album through generative tools as a flesh-and-blood musician and no badge appears. Release a record played entirely by humans under an invented persona and one does.
Spotify has no system today for detecting AI-generated music in the way Deezer does. What it offers is a channel for artists to disclose AI use, plus a “Verified by Spotify” badge launched to signal that someone is a real person. That’s disclosure, not detection.
So the badge settles a question about who is behind the profile while leaving the question most listeners actually have — was this made by a machine — largely unaddressed.
There’s an appeals process, and there needs to be
Artists will be notified when a “Likely AI Persona” badge lands on their profile, and they’ll be able to appeal it. That matters, because a false positive here is more than cosmetic. Losing your place in default recommendations hits a working musician’s income directly, and the word “photorealistic” is carrying enormous weight in a system tasked with assessing press photos.
The company also intends to ship a tool letting listeners report artists they suspect of using AI-generated identities. Crowd-sourced judgments about authenticity carry an obvious risk, and how that queue is handled will reveal far more about the policy than the badge’s design ever could.
When the rollout arrives in mid-September, watch for one thing above all: whether an artist already in your rotation ends up with a “Likely AI Persona” badge. That’s the real measure of how well Spotify’s human-plus-AI review is calibrated — and unlike the badge itself, it’s a test you can run for yourself.