The music world may look back at the fall of 2025 as the beginning of Peak AI Slop. As I write this, AI-generated tracks are now appearing on Spotify’s Global Viral 50 and multiple Billboard charts. In recent weeks, AI artists have fooled BBC announcers and slipped anti‑immigrant messages onto Spotify’s Dutch viral charts. This month Deezer estimated that roughly 50,000 fully AI-generated tracks are uploaded to its platform each day, about 34% of all new submissions. If that weren’t enough, Joe Rogan can’t stop raving about AI versions of 50 Cent tracks on his show.

Against this backdrop, the lack of urgency from major music streamers is getting harder to ignore. Spotify, the biggest and most algorithmically-driven service, is AI‑slop destination number one. In late September, the company said it had removed more than 75 million “spammy” tracks over the past year and updated its policies to ban unauthorized AI voice clones and deepfake‑style imitations; it also said it is “helping develop” an industry standard for disclosing AI in song credits.

What Spotify has not offered is a simple way for users to turn AI music off. In a recent Deezer/Ipsos survey, more than 80% of respondents said AI‑generated tracks should be clearly labeled, and 52% said they were uncomfortable not knowing whether what they were hearing was human or synthetic. TikTok, of all places, is preparing to give users more control over how much AI video content appears in their feeds. Why can’t the music streamers?

The technical challenges are real. Detection has to come first; a Deezer research team reports that its open‑source AI‑music detector can reach around 99.8% accuracy in test. But the company also stresses that detection is a cat‑and‑mouse game as new models learn to evade today’s acoustic fingerprints. There’s also nuance: what about songs with AI‑assisted production but human vocals? Or human‑written lyrics set to AI‑generated arrangements, as is the case with Xania Monet?

Perhaps more important, the major services are largely framing the issue as fraud rather than AI. Fraud — schemes to inflate plays and royalties through playlist manipulation, bots, click farms, and fake accounts — is something they’ve dealt with for more than a decade. AI, meanwhile, is a strategic pillar they’re talking up to shareholders and using across the product. So it’s both convenient and not entirely wrong for them to say this isn’t about AI music, it’s about people abusing the system.

But AI makes abuse exponentially easier. Five years ago, manipulators needed to find niches to fill, and so-called “fake artists” were often real musicians doing work-for-hire. AI lets anyone make music at scale, and dramatically expands the possibilities: instead of hiring bedroom producers to crank out hours of, say, instrumental chill music, you can now chase hits for next to nothing.

In the case of Breaking Rust, they likely bought their way onto Billboard’s Country Digital Song Sales chart, then used that chart position to grab attention and, along with coordinated TikTok campaigns and other tactics, gin up streams on Spotify’s Viral charts. Once your songs hit the top of Spotify’s Viral charts, the hope is that enough people are listening or hate-listening organically to keep the plays coming. (It’s likely not a coincidence that the one person linked to Breaking Rust, as we reported in the last issue of Stems, has also worked for an independent streaming‑promotion company.)

I’m sure some AI music enthusiasts would argue that popularity is the be-all, end-all: if a song reaches the top of a chart without blatant fraud, it belongs there. But most music fans want to know when they’re listening to a human.

That’s going to require distinct lanes for AI-generated music and human-made music. “You don’t confuse a TikTok clip with a Netflix series because they live in different lanes,” said Mark Mulligan of MiDia Research, "That’s where we need to get with AI.”

I’d love to hear your thoughts on this. Hit me at [email protected].