French music service Deezer reported that AI-generated tracks now account for 44% of all new daily uploads. Approximately 75,000 such compositions are added to the platform each day, totaling over 2 million per month. 

However, despite the volume, these songs are rarely listened to, making up only 1-3% of total streams. 

AI and Fraud

The gap between uploads and actual listens is attributed to fraud. 

Deezer's own technology identified 85% of AI track plays as artificial. Consequently, the platform excluded these from its royalty payment system.

The company launched Deezer AI Detection in January 2025, becoming the first service to label content generated by neural networks. The tool claims an accuracy rate of 99.8%. Over the past year, it has identified more than 13.4 million AI tracks in the platform's catalog. 

“AI music has ceased to be marginal. With the rise in daily uploads, we urge the entire industry to unite to protect artists' rights and ensure transparency for fans,” said Deezer CEO Alexis Lanturnier.

As of April 20, the platform also stopped offering high-quality versions of AI-generated compositions. 

People Can't Hear the Difference

Despite efforts to identify AI tracks, listeners struggle to distinguish them from human-made music. A Deezer study involving 9,000 participants from eight countries showed that 97% of users could not tell which compositions were created by humans and which by neural networks. 

Moreover, 80% of respondents supported mandatory labeling of such content for transparency.

Additionally, 73% of streaming listeners want to know if the service recommends entirely AI-generated music. 52% believe these tracks should not compete in the main charts alongside human-made music.

AI Bots Stealing Musicians' Names 

Previously, journalists from The Guardian reported on jazz pianist Jason Moran, who discovered an album released under his name on Spotify. The musician does not use the service and has not uploaded his music there. 

After an investigation, he found a profile on Spotify with his name and several albums from a label he had previously worked with. The account also featured an unfamiliar indie-pop EP. 

“There isn’t even a pianist on this entire recording. It doesn’t resemble what I do at all,” Moran commented. 

It turned out that AI bots were masquerading as real artists. Spotify acknowledged the issue. In September of last year, the service reported that it had removed over 75 million "spam tracks" in 12 months. At that time, the company announced enhanced protections for musicians, including stricter rules against impersonation.

Notably, in February 2026, Google released the AI model Lyria 3 for music generation. This neural network allows users to create audio using text prompts, photos, or videos.