Spotify faces scrutiny over AI music as third-party trackers fill transparency gap

The issue has raised concerns among artists and listeners about transparency, copyright and the impact of AI-generated music on streaming royalties. Unlike some competitors, including Deezer, Spotify has yet to introduce a widely available system that clearly identifies AI-generated or AI-assisted tracks.

One tracker, SoullessMusic.com, describes itself as a database tracking “AI artists hiding on Spotify”. It uses automated analysis to identify music it considers likely to have been generated using AI. The site has, for example, classified the AI-assisted R&B project Slime Dot as “almost certainly AI”, citing the volume of music released within a short period.

“I didn’t know what was AI and what’s not,” said Graeme Fulton, the creator of SoullessMusic, in comments to 404 Media. He said he became interested in the issue after seeing artists raise concerns on social media about AI projects allegedly copying their music and identities.

Another service, SlopTracker, allows users to submit Spotify track links for analysis. It has classified Slime Dot's song ‘Fully’ as 95% likely to have been generated by Suno, an AI music-generation platform.

However, the reliability of such third-party detection systems remains contested. Both services use AI and automated tools to assess music, which can produce false positives. SlopTracker previously identified the AI-assisted project Georgia Phantom as likely AI-generated but removed it after receiving a cease-and-desist notice. SoullessMusic continues to list the project, assigning it a 99% probability of being “almost certainly AI”.

The debate is not necessarily about whether musicians should use generative AI tools such as Suno and Udio, but about whether listeners and rightsholders should be informed when such technology has been used.

AI music detection systems can analyse characteristics associated with the technology, including vocoder artefacts and other digital signatures. Some systems use so-called “fakeprints” – machine-readable patterns associated with AI-generated audio – while neural fingerprinting can be used to identify musical material derived from existing recordings.

Detection tools can combine these methods with statistical and structural analysis to produce confidence scores rather than definitive conclusions. Their accuracy can also be affected by developments in AI technology, as well as techniques such as remastering, equalisation and other alterations that may obscure identifying characteristics.

The risk of incorrectly identifying human-made music as AI-generated remains a concern, particularly as detection methods continue to evolve alongside generative AI systems.

Spotify told 404 Media that it was taking a “layered approach” involving enforcement, artist controls and transparency.

“Over the past year, we’ve introduced policies targeting harmful AI-related behaviours like spam, impersonation, and deceptive content, alongside new tools that give artists more control and listeners more context,” the company said.

Spotify pointed to features including AI Credits, which allow artists to disclose how AI was used in their music, as well as Verified by Spotify and Artist Profile Protection.

The company said it would continue developing these measures. However, Artist Profile Protection requires artists to opt in, while AI Credits depend on artists voluntarily disclosing their use of AI.

The lack of a comprehensive labelling system has become more significant as AI-generated music grows. Deezer, which has invested in identifying and labelling AI-generated content, recently reported that almost half of new music uploaded to its platform was AI-generated and said it had begun removing certain AI-generated tracks from royalty payments.

SlopTracker estimates that AI-generated music appearing on Spotify’s official curated playlists is diverting revenue from human artists, although the methodology behind its financial estimates has not been independently verified.

The growing use of AI-generated music is therefore placing greater pressure on streaming platforms to improve transparency. While third-party databases are attempting to identify AI-generated content, questions remain over the accuracy of their methods and whether responsibility for disclosure should ultimately rest with artists, labels or the platforms distributing the music.