Deezer Lifts the Veil on AI-Created Music | Festina Lente - Your leading source of AI news | Turtles AI

Deezer introduces “AI generated content” labels on albums with artificially generated tracks, excludes them from automatic discovery and blocks royalty payments related to fraudulent streams. The move combats an 18% daily invasion of AI uploads.

Key Points:

• 18% of songs uploaded daily to Deezer are AI-generated
• About 70% of streams on these tracks are fake, powered by bots
• Deezer labels, excludes from playlists and blocks payments for these tracks
• Two patents filed in December 2024 define digital signatures of synthetic content

Deezer, a French platform known for its focus on transparency and copyright, announced a significant update to its features on Friday: all albums that include tracks generated entirely by AI will now be clearly labeled. This innovation is part of a strategy aimed at reducing streaming fraud, given that – despite representing only 0.5% of total streaming – these synthetic contents show suspicious behavior: 70% of their listening would be the work of bots, aimed exclusively at collecting royalties in a fraudulent manner.

The measure involves a triple intervention: the songs generated by AI are identified and reported via a visual badge, eliminated from indexing on editorial playlists and algorithmic recommendations, and the revenues deriving from suspicious streaming are excluded from the payment flow to the creators. This strategy is made possible thanks to an internal automatic detection system, based on algorithms that use the same music generators to identify digital patterns typical of synthetic songs; Deezer updates these tools daily to adapt them to the innovations of the opposing AI.

The volume of generated songs is growing rapidly: from around 10,000 daily uploads in January (10% of the total) it went to 20,000 in June (18%). While representing a negligible share of streams, this content dilutes the catalog and fuels fraudulent activities that exploit the pro-rata pay-per-listen model, already criticized for favoring automated streaming systems.

In December 2024, Deezer filed two patent applications for its AI Detection technology, which is based on the search for unique signatures to distinguish synthetic from authentic content. This tool, operational since January, allows to recognize and block 100% of tracks created with popular generators such as Suno and Udio, excluding them from recommendations and monetization flows.

The phenomenon is part of a broader scenario: major labels such as Universal, Warner and Sony are in talks to license their catalogs to AI music generators, such as Suno and Udio, to resolve ongoing legal disputes. In the meantime, Deezer – with over 10 million users – is stepping up its fight against manipulation, claiming an innovative approach: countering the same abuse generated by AI with AI tools.

The academic context confirms that detecting music deepfakes, while technically possible with accuracy close to 99%, requires robust systems against the continuous evolution and sophistication of AI.

In this context, Deezer is a pioneer in creating a transparent infrastructure to monitor and manage AI-generated music, with the aim of protecting users, legitimate artists and fair compensation models.

This approach lays the foundation for a more conscious coexistence between human creation and automatic generation, reflecting a balance between technological innovation and the protection of rights.