ERASYOpen App

AI Music

Spotify AI Music Policy 2026: What's Allowed, What Isn't

By Cara Whelan··12 min read
Spotify AI Music Policy 2026: What's Allowed, What Isn't — Erasy

Every few weeks someone tells us Spotify has banned AI music, and every time it is wrong — the Spotify AI music policy has never prohibited AI-assisted tracks, and Spotify says outright that it does not down-rank them for being declared. What changed in 2026 is quieter and costs more: a new AI Persona badge, deploying from mid-September 2026, pulls tagged profiles out of editorial and algorithmic recommendations by default. That is not a ban. For an independent release it is worse than one, and we have read every primary announcement behind it rather than the coverage of them.

This is the whole policy in one place: the four things Spotify actually enforces, which of them apply to a track and which to an artist identity, what the September badge does to your reach, and the six-step pass we run before submitting an AI-assisted release. Every figure is dated to the announcement that published it.

Key takeaways

AI music is allowed. Spotify's own wording: the policy is not about "punishing artists who use AI responsibly", and "all music is treated equally, regardless of the tools used to make it".

The AI Persona badge lands mid-September 2026. Announced 11 August 2026. It marks an artist identity, not a track, and shows on the profile, in search and on track rows.

The badge costs you discovery, not revenue. Spotify will not include AI Personas in editorial or algorithmic recommendations by default — only listeners who follow you still get served.

Enforcement targets behaviour, not generation. Over 75 million spammy tracks removed in twelve months for mass uploads, duplicates, metadata SEO hacks and artificially short tracks.

Declaring AI is the low-risk move. The DDEX AI credits beta went live 16 April 2026, and Spotify does not rely on self-disclosure alone — so the choice is a badge you declared or one you did not.

An artist profile card carrying a single badge marker, with the recommendation pathways nearest the badge fading out while the rest run on — Spotify's AI Persona badge and what it costs

The short answer, and the part it hides

You can release AI-generated music on Spotify. You can earn royalties on it. You do not need permission, a special account, or a human co-writer. If that were the whole answer this article would be three sentences long.

The part that matters is that Spotify runs four separate systems that an AI release can collide with, and only one of them is about AI at all. They were built at different times, for different reasons, and they punish different things. Treating them as one "AI policy" is why people end up surprised.

SystemWhat it targetsWhat it does to youLive since
Music spam filterMass uploads, duplicates, metadata SEO hacks, artificially short tracksUploader tagged, tracks stop being recommended; removal in volume casesRolling out from autumn 2025
Impersonation policyUnauthorised vocal impersonation of a real artist, fraudulent uploads to official profilesTrack removed; formal claims route for the impersonated artist25 September 2025
AI credits (DDEX)Nothing — it is a disclosure field, not an enforcement systemShows an AI tag in Song Credits. Spotify states it is not used to down-rankBeta, 16 April 2026
AI Persona badgeArtist identities that appear to be AI-generated rather than a real personBadge on profile, search and track rows; excluded from editorial and algorithmic recommendations by defaultDeploying from mid-September 2026

Read that table column by column and the shape of the policy appears. Three of the four systems care about what you do — how many files you upload, whose voice you used, what you declared. Only the fourth cares about what you are, and it is the newest. That is the change worth planning around.

Spotify's four enforcement systems side by side: the music spam filter live from autumn 2025, the impersonation policy from 25 September 2025, AI credits via DDEX in beta from 16 April 2026, and the AI Persona badge from mid-September 2026
Four systems, four trigger conditions · only the fourth is about AI

The AI Persona badge: what actually lands in September

On 11 August 2026 Spotify announced a label for AI-generated artist identities. In its own framing the badge signals that an artist's identity may be AI-generated and does not represent a real person, and it renders in three places at once: the artist profile, search results, and the track rows where your music appears. Deployment begins in mid-September 2026.

There are two routes to getting one. Artists can declare an AI Persona themselves through Spotify for Artists, which is available now. And Spotify's review team applies the badge independently to profiles whose public identity appears to represent a photorealistic AI-generated person, starting with artists who have met defined audience thresholds — the announcement does not publish the numbers, only that the thresholds cover the vast majority of artist profiles listeners actually visit. Flagged artists are notified, and can appeal or self-disclose.

11 Aug
2026 announcement
Mid-Sept
2026 rollout begins
3
Surfaces the badge appears on
0
Tracks removed by it

The sentence that decides whether this matters to you is the one about recommendations. By default, Spotify will not include AI Personas in any editorial or algorithmic recommendations unless a listener already follows the profile. Nothing is taken down. Streams still pay at the same rate. You are simply not in Discover Weekly, Release Radar, autoplay or editorial consideration any more.

For an established artist with a following, that is survivable. For a new project with no followers, algorithmic recommendation is the distribution — and being outside it means your release reaches the people you personally send it to and nobody else. Read that as the real price of the badge, because the badge itself is cosmetic and the exclusion is not.

What the badge does not do
  • Remove your tracks or your profile
  • Change your royalty rate or withhold payouts
  • Apply to individual songs — it marks the artist identity
  • Follow from declaring AI in your song credits
  • Trigger automatically on AI-assisted production
What it does do
  • Label the profile in three places listeners see
  • Remove you from editorial and algorithmic recommendation by default
  • Apply after Spotify review even if you never declared
  • Arrive with a notification and an appeal route
  • Leave followers' feeds and direct links working normally

One distinction is worth holding onto because almost every summary blurs it: an AI Persona is a fabricated artist — a name, a face, a biography that no person stands behind. A real person releasing music they made with Suno is not an AI Persona, and nothing in the announcement suggests they are. If you release under your own name with your own photograph, this policy is not aimed at you.

What Spotify's AI Persona badge does and does not do, with the announcement on 11 August 2026 and deployment beginning mid-September 2026
Spotify newsroom, 11 August 2026
Spotify's For the Record post introducing a new label for AI-generated artist identities, dated August 11 2026, showing the AI Persona badge on an artist profile
Source: Spotify For the Record, 11 August 2026

The music spam filter, and why it catches non-AI artists too

This is the system that has actually removed music at scale, and it predates the badge by a year. Spotify reported removing over 75 million spammy tracks in the twelve months to September 2025 — a number worth sitting with, because it is roughly the size of the entire catalogue a decade earlier.

The filter identifies uploaders and tracks engaging in four specific tactics, tags them, and stops recommending them: mass uploads, duplicate uploads, search-engine-optimisation hacks in metadata, and artificially short tracks designed to farm per-stream payouts. Notice that none of those four words is "AI". A human producer uploading the same instrumental forty times under forty titles trips this. A careful AI release does not.

Deezer took roughly 90,000 fully AI-generated tracks a day in July 2026, over 50 per cent of daily uploads, for 1 to 3 per cent of streams, alongside more than 75 million spammy tracks removed by Spotify in twelve months
Deezer’s upload figures, July 2026 · Spotify’s removal figure, September 2025

The volume context explains the machinery. Deezer, which publishes more detail than anyone, said in July 2026 that fully AI-generated music had passed 50% of its daily uploads at roughly 90,000 tracks a day, while accounting for between 1 and 3% of actual streams. Deezer's newsroom post is the clearest public statement of the problem every platform is solving: an enormous inbound pile, very little of it listened to, and a meaningful slice of it fraud.

Your job is to not look like that pile. In practice that means one artist name rather than five, releases spaced like a person makes them, titles and credits written by a human, and no thirty-second filler tracks padding an album. We go through the upload-side specifics in our guide to distributing AI music without getting flagged.

Impersonation: the one true prohibition

If there is a hard rule in the Spotify AI music policy, this is it. Vocal impersonation is only allowed in music on Spotify when the impersonated artist has authorised the usage. That is the published wording, from the 25 September 2025 announcement, and it applies whether the impersonation came from a voice model or a gifted human mimic.

The same policy covers fraudulent uploads to an artist's official profile — the trick where someone delivers a track under an existing artist's name so it lands on their page. Spotify said it is testing prevention upstream with distributors, which is why some distributors now hold releases that name-match an established act.

The practical consequence for AI creators is narrow but absolute: an AI cover in a recognisable singer's voice has no compliant path onto Spotify without documented authorisation from the rights holder, and no amount of audio cleanup changes that, because the violation is the voice rather than the file. We cover what remains possible in our guide to AI cover songs. Everything else in this article is about tracks that are yours to release.

AI credits, DDEX, and whether honesty costs you

The question we get more than any other is whether ticking the AI box at upload is an own goal. The evidence says no, and it is unusually explicit for a platform policy.

Spotify helped develop the industry disclosure standard through DDEX, the body that defines how music metadata moves between distributors and services. The standard lets a delivery state where and how AI played a role — AI-generated vocals, instrumentation, or post-production — as separate fields rather than one blunt flag. Spotify's AI credits beta went live on 16 April 2026, surfacing those credits in the Song Credits panel on mobile.

On the consequence, Spotify's own announcement says the work is "not about punishing artists who use AI responsibly or down-ranking tracks for disclosing information about how they were made", and that "all music is treated equally, regardless of the tools used to make it". You are entitled to be sceptical of a platform describing its own fairness. What you cannot claim is that the policy is secret or ambiguous.

The asymmetry that should decide it for you

Not declaring does not make a track undetectable. It moves the decision from you to a classifier — Spotify's, Deezer's, or your distributor's — which reaches its own conclusion without your input and without telling you.

A declared AI track carries a credit. An undeclared one later identified as AI carries a credibility problem, and on Spotify it now risks a badge applied by review rather than by choice.

Spotify against the platforms it is compared to

Spotify is not the strictest service on AI music, which surprises people who only read the headlines about it. Deezer moved earlier and enforces against the audio itself rather than the identity behind it.

PlatformPosition on AI musicWhat it labelsRecommendation impact
SpotifyAllowed; enforcement aimed at spam, impersonation and AI identitiesAI Persona badge on the artist, from mid-Sept 2026; AI credits per track since Apr 2026Badged personas excluded from editorial and algorithmic recommendation by default
DeezerAllowed; own AI detection running since January 2025Albums containing fully AI-generated tracks tagged for listeners since June 2026Detected fully-AI tracks excluded from algorithmic and editorial recommendation
Apple MusicAllowed; AI Transparency Tags announced to providers 4 March 2026Self-declared tags across Artwork, Track, Composition and Music Video — optional now, required on future deliveriesNo published AI-specific recommendation exclusion
YouTubeAllowed with disclosure of synthetic or altered realistic contentDisclosure label on the watch page where requiredMonetisation governed by Partner Program and originality rules, not an AI flag
Distributors (DistroKid, TuneCore, CD Baby)Accept AI music if you hold the rights and declare itAI declaration field passed on in the DDEX deliveryScreening happens before the platforms ever see the file

Apple is the interesting third case, because it went the opposite way to Deezer. Its AI Transparency Tags, announced to content providers on 4 March 2026, split disclosure into four fields — artwork, track, composition and music video — and Apple says it defers to content providers to determine what qualifies as AI-generated content. The tags are optional now and required on future deliveries, with no independent verification behind them. That is disclosure without detection, which is the mirror image of Deezer's approach and a different risk profile again from Spotify's review team.

Two things follow from that table. First, the gate you meet earliest is your distributor's, not Spotify's — a file that fails an automated screen at submission never becomes a Spotify policy question at all. Second, Deezer's model is the one to watch, because exclusion triggered by detection can reach you whatever you declared, while Spotify's badge at least comes with a notification and an appeal.

If you want the detail on how those detection systems reach a verdict and why two of them routinely disagree about the same file, we pulled that apart in our piece on what AI music detectors actually see.

Where Spotify, Deezer, Apple Music, YouTube and the distributors each stand on AI music, what each one labels and whether it affects recommendation
Deezer enforces against the audio · Spotify against the identity

The pass we run before an AI release goes to Spotify

Policy compliance and passing the screen are two different jobs, and this order handles both. Nothing here takes specialist software except step three.

01
Confirm the rights

Generate on a paid tier so the commercial licence attaches to the audio, and keep the receipt. Distributors ask for proof, and it is the one rejection category with an easy answer.

02
Decide the identity question

Releasing under your own name and face? The Persona badge is not aimed at you. Building a fictional artist? Self-declare it and keep the framing and the appeal route yours.

03
Clean the file

Repair generation artifacts, clear embedded provenance metadata and generation markers, and do it before you master rather than after so you are not re-colouring balanced audio.

04
Master to platform spec

About -14 LUFS integrated with true peak at or below -1 dBTP. Spotify turns loud masters down rather than rewarding them, and AI exports usually arrive too quiet rather than too loud.

05
Write metadata a human would write

One artist name, no keyword stuffing in titles, real credits, no near-duplicate versions of the same track. This is what the spam filter reads.

06
Declare the AI involvement

Tick your distributor's AI field and select which elements applied. It travels as a DDEX credit and, on Spotify's own statement, does not cost you ranking.

Steps three and four are the ones people skip, and they are the ones that decide whether your distributor forwards the release at all. The full version, timed step by step, is in our twelve-step cleaning checklist, and the loudness numbers per platform are in our LUFS targets guide.

Before the distributor screen, not after

Clear the file once. Let the policy be the only question.

Undetectr handles the audio half of this list — generation artifacts, embedded watermarks, C2PA provenance metadata and noise-floor patterns — in a single pass, on a one-time Lifetime plan rather than a subscription.

The six-step pass before an AI release goes to Spotify: confirm the rights, decide the identity question, clean the file, master to about -14 LUFS with true peak at or below -1 dBTP, write human metadata, and declare the AI involvement
Steps three and four are the ones people skip

Four things people get wrong about this policy

"Spotify banned AI music in 2025." It did not. It removed 75 million spammy tracks and published an impersonation policy in the same week, and the two got compressed into one headline. The announcement that produced those numbers says in writing that all music is treated equally regardless of the tools used to make it.

"The AI badge will be slapped on my song." The badge marks an artist identity, not a track. Declaring AI involvement in a song's credits is a separate system that has existed since April 2026 and carries no recommendation penalty.

"If I do not declare, nobody knows." Spotify said explicitly that it will not rely on self-disclosure alone and reviews profiles itself. Deezer has been detecting from the audio since January 2025. Non-declaration is not invisibility, it is forfeiting your say.

"Stripping the watermark solves the policy problem." It solves a different problem — the automated screen between you and the platform. No amount of audio work makes an unauthorised voice clone compliant, or a forty-track duplicate upload look like a human release. Fix the audio because the distributor gate is real; fix the behaviour because that is what the policy reads.

And a note on where this is heading. Spotify says listeners will eventually be able to report suspected AI Personas, which turns enforcement from a review queue into a crowd signal. If you are building a project you intend to run for years, the version of it that survives that shift is the one that was honest from the start.

Frequently asked questions

The Erasy wordmark in orange lower-case lettering on black, shown as the byline mark for Cara Whelan, Policy Editor at Erasy

Cara Whelan — Policy Editor, Erasy

Cara covers platform policy for Erasy — pricing changes, terms updates and the rules that decide what you can actually do with the music you generate. They work from primary announcements and FAQs rather than the coverage of them, and date every figure so you know when it was last checked.

Does Spotify allow AI-generated music in 2026?+
Yes. Spotify has never banned AI-assisted or AI-generated music, and its September 2025 policy announcement said so directly: the approach is "not about punishing artists who use AI responsibly or down-ranking tracks for disclosing information about how they were made", and "all music is treated equally, regardless of the tools used to make it". What Spotify enforces against is behaviour and identity — spam upload tactics, unauthorised vocal impersonation, and from September 2026 undisclosed AI artist personas. Being made with Suno is not itself a violation.
What is Spotify's AI Persona badge and when does it appear?+
It is a label announced on 11 August 2026 that marks a profile whose artist identity may be AI-generated and does not represent a real person. Spotify said deployment begins in mid-September 2026, and the badge shows on the artist profile, in search results and on track rows. Artists can self-declare through Spotify for Artists, and Spotify's own review team also applies it to profiles that appear to present photorealistic AI-generated identities, starting with artists above defined audience thresholds. Flagged artists are notified and can appeal or self-disclose.
Does the AI Persona badge stop you earning on Spotify?+
Not directly — it is a label, not a takedown, and streams still pay. The cost is distribution: Spotify said that by default it will not include AI Personas in editorial or algorithmic recommendations unless a listener follows the profile. For most independent releases discovery is the entire growth engine, so being outside the recommendation system is a serious commercial hit even though nothing is removed. It applies to the artist identity, not to individual tracks.
Will disclosing AI in my credits get my track down-ranked?+
Spotify states plainly that it does not down-rank tracks for disclosing how they were made, and it helped develop the DDEX standard that carries the disclosure. The AI credits beta went live on 16 April 2026, showing in Song Credits on mobile which elements — vocals, instrumentation, post-production — involved AI. Declaring at upload is the low-risk move: it puts one accurate flag in the metadata chain, and the alternative is not invisibility but a classifier deciding for you without your input.
What actually gets AI music removed from Spotify?+
Spam behaviour and impersonation, not generation. Spotify removed over 75 million spammy tracks in the twelve months to September 2025, targeting mass uploads, duplicate uploads, SEO manipulation in metadata and artificially short tracks used to farm royalty payouts. Separately, unauthorised vocal impersonation of a real artist is a policy violation regardless of whether AI was involved. A single well-made AI-assisted track with honest metadata does not meet either test.
How is Spotify different from Deezer on AI music?+
Deezer is stricter and moved first. It has run its own AI detection since January 2025, and in June 2026 it became the first streaming service to tag albums containing fully AI-generated tracks for listeners; tagged tracks are excluded from algorithmic recommendations and editorial playlists automatically. Spotify's badge is applied to the artist identity rather than detected per track, and its recommendation exclusion follows the badge. The practical difference is that Deezer's exclusion is triggered by detection of the audio, so it can hit you without any declaration at all.
Do I need to change my tracks to comply with Spotify's policy?+
Compliance is mostly metadata and behaviour rather than audio. Own the rights, declare AI involvement through your distributor's AI field, use one artist name, avoid duplicate and near-duplicate uploads, and do not clone a real singer's voice without written authorisation. The audio work — artifact repair, embedded provenance markers, loudness — matters for a different reason: it is what decides whether your distributor's automated screen passes the file to Spotify at all.
Should I hide that my artist project is AI?+
We would not, and the mechanics are against you. Spotify said it will not rely on self-disclosure alone and reviews profiles independently, so the realistic outcomes are a badge you declared or a badge you did not. Self-declaring keeps the appeal route and the framing yours; being tagged after review means the label arrives with no context and no warning. Neither outcome removes your music, and both carry the same recommendation exclusion.

Disclosure: Erasy is an independent guide to AI music cleanup, and Undetectr is the tool we use and link to. The AI Persona badge details, rollout timing and recommendation treatment are from Spotify's newsroom announcement of 11 August 2026; the spam-filter, impersonation and DDEX AI-credits details and the 75 million figure are from Spotify's announcement of 25 September 2025. Deezer upload and stream figures are Deezer's own, published July 2026. Apple's Transparency Tags detail is from its 4 March 2026 provider announcement, as reported by Music Business Worldwide. Platform and distributor policies reflect published rules as of 6 September 2026 and change without notice — check the current terms before you submit.