Press play, and you may now get a straight answer about whether a human made the song.
YouTube Music started labeling tracks made primarily with generative AI on Oct. 14, covering generated compositions and vocals across its app and the main YouTube app. Deezer announced a filter, rolling out worldwide by the end of 2026, that lets users hide fully AI-generated tracks, albums, and artists from search results. Deezer already excludes detected AI tracks from its algorithmic and editorial recommendations; the new toggle extends that control to searches users run themselves. Both moves follow Spotify's AI Persona badges, launched in September, which flag photorealistic AI artist profiles without specifying how the music itself was made.
This matters because detection, not disclosure, is the real bottleneck. A Deezer/Ipsos survey found 97 percent of listeners couldn't tell two AI songs from a human one by ear, yet 80 percent wanted AI music labeled anyway. Deezer says it receives nearly 90,000 fully AI-generated tracks a day, and that up to 85 percent of streams on those tracks in 2025 were fraudulent - the same royalty-pool exploit that got North Carolina musician Michael Smith an 18-month prison sentence on Oct. 6 for an $8 million streaming fraud scheme.
None of these systems agree on what they're even disclosing. Spotify's badge addresses artist identity, Deezer and YouTube target how a track was built, and Apple's reported Transparency Tags rely on labels self-reporting the details. A song with no label could mean a human made it, or just that nobody's detection caught it yet. Until platforms compare notes, listeners are getting four different partial answers to one question.