Spotify lets 1000 flowers bloom, but tells you which ones are plastic.

Nobody asked whether auto-tune was cheating when T-Pain made it a genre. Nobody held a parliamentary inquiry into drum machines before they rewrote pop music. We are asking those questions about AI music now, and Spotify has quietly offered its answer: let it exist, label it, and let listeners decide.

Spotify is applying social pressure toward transparency, not a ban — and that distinction matters

The mechanics of what Spotify has announced are modest enough. Verified artists get a green checkmark. A new beta feature, AI credits, allows musicians to disclose where and how AI touched their work: the instrumentation, the vocals, the production. Importantly, that disclosure is voluntary. Spotify's global head of artists, Sam Duboff, concedes some tracks will slip through. The platform is not a regulator, and it is not pretending to be one.

What it is doing is something structurally useful: applying social pressure toward transparency without imposing a ban. That distinction matters. A ban would feel decisive, but it would be a solution borrowed from the wrong problem. The question is not whether AI should exist in music. It already does, invisibly, in pitch correction, beat quantisation, mastering algorithms, and sample libraries that approximate real orchestras. The question is whether listeners can tell what they are hearing and whether they have been misled about its origins.

Slop that nobody streams earns nothing and dilutes nothing. The economic threat it poses to working musicians is real but much smaller than the discourse implies.

The AI music panic is outrunning the actual numbers

On that score, the numbers are clarifying. Entirely AI-generated music, the industrially produced, low-human-input variety that genuinely worries critics, accounts for well under one per cent of consumption on Spotify. Spotify removed 75 million spammy tracks in twelve months, which sounds alarming until you note that those tracks were not actually being listened to. The platform's royalty pool is distributed by stream share: if a track gets one per cent of plays, it gets one per cent of the revenue. Slop that nobody streams earns nothing and dilutes nothing. The economic threat it poses to working musicians is real but much smaller than the discourse implies.

The more culturally live question is the Josh Fawaz case, a cover of Madonna's Like a Prayer that has been streamed more than 37 million times and sits atop Australia's national airplay chart amid conjecture about AI involvement. Spotify has not weighed in on that specific track. What it has said is that AI remixes and covers tend to escape scrutiny because they are made off-platform, go viral on social media, and arrive on streaming services already carrying momentum. The new system, where artists choose whether their songs can be remixed and where all plays are credited, is designed to close that loop. Whether it will is an open question; the system depends on good faith disclosure at the point of upload.

Every new music tool looked like a threat before it became a genre

Here is where the auto-tune comparison bites. When a new tool arrives in music, the first thing it produces is usually embarrassing. Drum machines gave us novelty records before they gave us hip-hop. Auto-tune gave us ringtone pop before it gave us Kanye West's 808s & Heartbreak. Digital sampling gave us lawsuits before it gave us a new vernacular for how music references itself. The first wave of any tool tends to reveal its limits more than its possibilities. The artists who matter are usually the ones who figure out what the tool can do that nothing else can, and that discovery takes time.

Banning or aggressively restricting AI in music now would mean foreclosing that discovery before it happens. It would also be largely performative: the tools exist, they are cheap, they are already in use, and a streaming platform's content policy cannot uninvent them. What a platform can do is shape the norms around them. Transparency about AI use is a norm worth establishing. It preserves listener trust, maintains the signal value of human authorship for artists who want it, and leaves space for the interesting hybrid work that has not been made yet.

Spotify's framing is commercially self-interested, as all platform policy is. A service people trust is a service people keep paying for. But self-interest and good policy are not mutually exclusive. The structure here is sound: label the tool use, remove the spam, let the market sort genuine creative work from industrial content generation, and trust that listeners are capable of deciding what they value.

They almost always are.

Sources

ABC News — Spotify removes 'AI slop' and seeks transparency from musicians who use AI

Frequently Asked Questions

How does Spotify's new AI music labelling system work?
Verified artists on Spotify can now voluntarily disclose how artificial intelligence was used in their music through a feature called AI credits, covering elements like vocals, instrumentation, and production. Artists also receive a green checkmark confirming their verified status. Disclosure is not mandatory, and Spotify has acknowledged that some AI-assisted tracks will not be declared.

Is AI-generated music taking over Spotify?
Not by the numbers. Purely AI-generated tracks account for well under one per cent of actual consumption on the platform. Spotify did remove 75 million spammy tracks in twelve months, but those tracks were not being listened to — and under Spotify's royalty model, tracks that don't get streamed don't earn revenue or dilute the pool available to human artists.

Why doesn't Spotify just ban AI music?
Because a ban would be largely performative. The tools already exist, are cheap, and are in widespread use — a streaming platform's content policy cannot remove them from the world. What Spotify can do is shape the norms around AI use, and a transparency-and-labelling approach preserves listener trust while leaving room for legitimate creative experimentation.

Do AI music tools hurt musicians financially?
The threat is real but smaller than the public debate implies. Spotify distributes royalties proportionally to stream share, so AI-generated tracks that attract no listeners earn nothing and take nothing from working musicians. The genuine risk is AI content that is actively consumed at scale — which, for now, remains a small fraction of total listening.

What is the Josh Fawaz AI music controversy?
A cover of Madonna's Like a Prayer attributed to Josh Fawaz has been streamed more than 37 million times and reached the top of Australia's national airplay chart, prompting public speculation about AI involvement in its production. Spotify has not commented on the specific track. The case illustrates a gap in the new labelling system: tracks that go viral on social media before arriving on streaming platforms may carry momentum that makes post-upload disclosure largely moot.