With Spotify still not labelling most AI-generated music, independent sites have started doing the job themselves, 404 Media reports. SoullessMusic keeps a database of artists it believes are AI, and SlopTracker lets anyone paste in a Spotify link or upload a file to get a probability that a track came from a generator.
The case that opens the report is Slime Dot, a Las Vegas R&B act with more than 100,000 monthly Spotify listeners who publicly denied being AI in a GQ interview. GQ later added a note conceding the act was an AI avatar. SoullessMusic lists Slime Dot as almost certainly AI, and SlopTracker rated one of its songs 95 percent likely to have been made with Suno. Another example is Qajar Jazz, a channel posting hours of Iranian jazz each week, which YouTube marks as AI in the small print while Spotify does not.
SoullessMusic founder Graeme Fulton says he started after seeing musicians complain that AI acts were copying their songs and look. His detector is open source, built on models such as SONICS and a vocoder fingerprint tool, and also checks metadata for traces of specific generators; he reviews every submission by hand. He admits false positives will happen, and the report points out that using AI to detect AI is flawed by design.
The money involved is estimated, not known. SoullessMusic puts the yearly earnings of the AI acts it tracks at about $5.7 million, with the top one, mikeeysmind, near $1.5 million. Deezer, which does tag AI uploads, said in April that 44 percent of its new tracks were AI-made.
After publication Spotify said it takes a layered approach: rules against spam, impersonation and deception, a Verified by Spotify badge, AI Credits that artists fill in themselves, which it says now arrive by the tens of thousands a day, and Artist Profile Protection. The reporter says the promised AI disclosures did not appear on any of the dozens of tracks reviewed, even though some generators embed watermarks Spotify could read.
Sources · 1