It feels like Spotify is trying to have its cake and eat it here. It says it’s pulling slop and spam from recommendations at the same time that it’s been adding genAI features to its platform. Spotify allows users to remix and cover songs with AI tools, for instance.
Some “artists” that are widely believed to be fake — such as Sienna Rose and The Velvet Sundown — have more than 100,000 monthly listeners on the platform. Another called Breaking Rust topped a Billboard digital downloads chart last year and currently has around 700,000 monthly listeners. A report in July stated that the top 10 most-listened-to AI artists on Spotify had raked in about $2.4 million, siphoning money away from the human musicians that Spotify claims it’s trying to promote.
The company is aware that a significant proportion of its users do not want to hear genAI garbage, but it only confirmed that it removed “slop and spam” tracks from recommendations, not necessarily the platform as a whole. Deleting all of that crap should be an easy decision if your aim is genuinely to put “human artistry first.”
Nor does Spotify offer users an easy way to block this gunk. This month, Deezer said it would give users an option to filter out genAI audio (I bristle at the idea of calling it music) entirely by the end of the year. Great! Bandcamp largely forbids genAI audio, while Tidal doesn’t pay royalties for tracks that are entirely AI generated.
It would be nice to see Spotify giving users a way to opt out of hearing AI-generated audio. However, a spokesperson told the BBC earlier this year its “priority is addressing harmful uses [of AI] like spam and impersonation, rather than trying to filter music based on how it was made.” Still, it could at least offer users a toggle to entirely block tracks identified as slop and spam, and a way to filter out those that have disclosed AI use in the credits.
Speaking of which, Spotify says artists are submitting tens of thousands of AI credits per day to clarify what role the technology played in their uploads. It added that users can see these credits in the Now Playing view on mobile and it’s adding more ways for them to “quickly and easily see when an artist has disclosed AI use as part of the song’s creative process.” Thousands of virtual artists now have an AI Persona label too, and tracks from these phonies won’t appear in editor-curated playlists or, by default, in your recommendations (unless you follow them, for instance).
Spotify claims its threshold of 1,000 streams before it pays royalties on a track is helping to stem the flood. “Tracks with indicators showing a high probability of being AI-generated are 10 times more likely to fall below the threshold, making it harder for bad actors to attempt to siphon royalties away from real artists by generating music at scale,” it said.
In addition, the company noted more than 300,000 human artists are now verified on the platform, covering more than 99 percent of those that users actually search for. Spotify is also expanding a tool that lets musicians approve what songs appear on their profile and list them as an artist. It will be available to all artists by the end of the year.