Spotify fights AI 'slop' uploads

Spotify is having an AI identity crisis. On one side, it is trying to clean up music recommendations that users say are getting clogged with fake artists and low-effort AI uploads. On the other, it just launched a new way to put AI-made audio directly into your library. That tension is the story here — Spotify wants more audio on the platform, but listeners want better signal, not more noise. ### What changed this week? On May 7, Spotify launched a beta called Personal Podcasts. The idea is simple: if you use an agent like OpenAI Codex, Claude Code, or OpenClaw on desktop, you can generate a private audio briefing and save it straight into Spotify. Think daily digests, study notes, or calendar rundowns — basically AI-made spoken audio that lives next to your music and podcasts. (newsroom.spotify.com) ### Why are users mad about music, then? Because a lot of them feel the recommendation system is already polluted. Android Police ran a hands-on test this week with third-party tools built to block likely AI artists on Spotify, and the whole premise only works because users think fake or mass-generated music is showing up often enough to distort discovery. The complaint is not “AI exists.” It is “I asked for new music and got content-farm filler.” (newsroom.spotify.com) ### What is Spotify’s fix? The main visible fix is the new Verified by Spotify badge. Spotify’s criteria center on real artist signals — sustained listener activity, policy compliance, and signs that an actual person or group is behind the profile, like linked socials, merch, or touring activity. Spotify says that at launch, more than 99% of artists users actively search for are verified, representing hundreds of thousands of artists. AI-generated artists are not currently allowed to apply. (androidpolice.com) ### Why doesn’t that solve it? Because verification is a proxy, not a detector. A missing badge does not prove an artist is fake, and a recommendation feed is full of acts people were not actively searching for in the first place. Spotify is also backing an industry standard for AI disclosures in music credits, but that relies on voluntary disclosure. Basically, if the bad actor never tells you the track was AI-made, the label system cannot rescue you. (androidpolice.com) ### Why are third-party blockers suddenly interesting? They do the thing platform tools still do not fully do — let users act on suspicion. The Android Police piece points to scripts and community lists that block artists associated with AI-generated music, which can make discovery feel cleaner fast. That is a scrappy workaround, not a real platform solution, but it shows where demand is. Users want a “show me fewer of these” switch. (androidpolice.com) ### How big could this get? Pretty big. Deezer said in January that it was seeing more than 60,000 AI-generated tracks uploaded per day — about 39% of daily intake — and that up to 85% of streams on AI-generated music were fraudulent. Deezer also says it tags AI tracks and removes them from algorithmic recommendations. That matters because it shows the problem is not theoretical and the stronger response is technically possible. (androidpolice.com) ### So why add more AI audio now? Because Spotify sees two different businesses. AI music spam is a trust problem. Personal AI briefings are a utility product. From Spotify’s point of view, a private study recap or morning briefing is not competing with artists for royalties in the same way a flood of fake songs does. But for listeners, the line is blurrier — it still feels like the app is opening the door wider to synthetic audio while trying to sweep synthetic music out of the hallway. (newsroom-deezer.com) ### What’s the bottom line? Spotify is not really “fighting AI” so much as trying to sort useful AI from junk AI. But until users get direct controls over recommendations, the cleanup will feel partial — and every new AI feature will make that contradiction more obvious. (newsroom.spotify.com)