Music streaming platform Deezer has revealed a striking figure: nearly 75,000 AI-generated songs are uploaded every day to the platform. This represents 44% of daily uploads, out of a total of more than 2 million (yes, you read that right) songs of this type uploaded per month.
Is Deezer doing anything to reduce this problem?
The platform says that, thanks to “industry-unique measures,” this type of content still represents a very small share: only 1% to 3% of total streams. In addition, the majority (85%) is detected as fraudulent and demonetized. Until June 2025, Deezer was the only music streaming service that identified whether content was artificial or not, and more than 13.4 million tracks were detected and labeled as AI-generated in 2025.
Collaboration with other platforms
According to CEO Alexis Lanternier, the detection technology has been in use since January 2025 and is available for licensing, which could help other platforms join the fight against this type of content.
But can people actually tell the difference between real and AI-generated music?
Deezer released a very interesting study showing that 97% of people were unable to identify any difference between AI-generated and human-created content. In addition, 80% of respondents agreed that all AI-generated music should be clearly labeled.
Risks to the music industry
According to a study conducted by CISAC and PMP Strategy, nearly 25% of creators’ revenue will be at risk by 2028, totaling €4 billion by that time.
What does the platform do when it identifies this type of content?
Once the system identifies a song as AI-generated, it is no longer recommended to users and is also removed from editorial playlists.
What do the studies say?
According to a survey conducted by the company itself across eight countries (USA, Canada, Brazil, United Kingdom, France, Netherlands, Germany, and Japan), 52% of people feel that 100% AI-generated music should not appear in charts alongside music created by humans.
What about Spotify?
In September 2025, Spotify released a statement saying that over the past 12 months it had removed more than 75 million tracks from the platform. In addition, the Swedish company also mentioned that it is taking other measures to reduce this type of problem:
1 – Rules on “impersonation” to prevent others from cloning artists’ voices. It is worth noting that in 2023, Spotify and other platforms removed a track that had cloned, without permission, the voices of Drake and The Weeknd.
2 – Spam music filters: detection and removal of tactics such as mass uploads, duplicate tracks, and attempts to manipulate search engine optimization (SEO), among others.
3 – AI content detection and disclosure system: Spotify makes it clear there is no binary way to identify content (AI or not AI), but rather many nuances in between. The system allows artists to indicate whether any part of a track was generated artificially, whether vocals, lyrics, or production.
A band that never existed: The Velvet Sundown
Imagine a typical psychedelic band with a strong 1970s vibe that quickly appears in users’ feeds and across multiple playlists, reaching hundreds of thousands of listeners in just a few weeks and accumulating millions of streams, with nearly 80,000 followers. It would be impressive—except for one thing: it wasn’t a real band. All the music was generated by AI. The account is still there, available for anyone to listen.
For now, Spotify appears to be much more permissive with this type of content. Will that continue in the future?
