AI Diffusion Models Take Over Music Generation As Deezer Reports A Boom In Fully AI-Generated Track Uploads

Ever since diffusion models burst onto the scene in 2022, they’ve reshaped what’s possible with Artificial Intelligence. First, they transformed visual media with platforms such as ChatGPT’s Dall-E 3, Midjourney, Stable Diffusion, and others, making image generators accessible.

Now, these models are now making noise—literally—in the music industry.

Fast forward to 2025, and the music scene is witnessing the same transformation as visual media did a few years ago. With the rise of music-generation tools powered by diffusion models, AI tools have moved from augmenting musicians to standalone creators.

This shift has prompted both excitement and concern, especially from music industry heavyweights like Deezer. So, what is happening with AI-generated music, and what revelations did Deezer make?

Let’s break it down!  

What Did Deezer Say About AI-generated Music?

The French music streaming service, Deezer, announced that over 20,000 fully AI-generated tracks were uploaded to its platform daily in April—nearly doubling the number from January this year. Such AI-generated songs now make up 18% of Deezer’s daily uploads, highlighting the widespread influence of generative AI-powered music.

The explosion in AI-generated tracks is largely driven by music diffusion models, such as Suno and Udio, which are capable of generating full-length songs using simple text prompts.

Aurelien Herault, the Chief Innovation Officer at Deezer, stated, “AI-generated content continues to flood streaming platforms like Deezer, and we see no sign of it slowing down… We need to approach the development with responsibility and care in order to safeguard the rights and revenues of artists.”


Hence, in January 2025, Deezer launched an AI detection tool to help identify and filter fully AI-generated music. While the tool doesn’t ban such tracks, it enables users to remove them from recommendation algorithms and editorial playlists.
 
According to Deezer, this policy protects human artists and preserves fairness in a rapidly evolving creative landscape. Users can still discover AI-generated music on Deezer—just not through algorithmic suggestions, as the streaming giant wants to preserve the discovery space for human-created music.

So, how does the tool identify fully AI-generated tracks? 

How Does Deezer’s AI Detection Tool Work?

Deezer’s tool currently detects AI tracks created using Suno and Udio models—two of the most popular tools in the generative music space.

It works by analyzing audio fingerprints and generation patterns. If a song is flagged as 100% AI-generated, it gets removed from algorithmic recommendations. However, the tool does not yet catch partially AI-generated tracks, accounting for a large portion of music being created today.

As per the U.S. Copyright Office, AI-assisted works can still qualify for copyright if there’s “sufficient human creativity” involved. While Deezer is the first major DSP (digital streaming platform) to publicly report purely AI music on its platform, it currently takes no regulatory action on the use of diffusion models.

So, what does this mean for AI-friendly artists and labels?  

Artists and Labels Push Back Against AI-generated Music

The increase in AI-generated tracks comes at a time when record labels, streaming platforms, and artists are pushing back against unauthorized use of their recordings to train AI models.

In late 2024, Universal Music Group, Warner Music Group, and Sony Music filed lawsuits against AI music generation platform Suno and Udio. With the allegations of mass copyright infringement, the labels claimed that the AI models were trained on their artists' songs without their consent.

Popular musicians such as Billie Eilish, Nicki Minaj, and Stevie Wonder also penned an open letter warning that AI-generated music threatens to “sabotage creativity” and “sideline human artists.”

So, what’s the divisive AI technology that has created such a rift in the music world?  

What Are Diffusion Models In AI?

Diffusion models work by generating waveforms from scratch, rather than stitching samples together, allowing for more organic, realistic, and stylized tracks.

Tools such as Udio and Suno allow users to describe the mood, genre, and lyrics, then generate polished audio tracks in minutes. This removes previous barriers to entry for amateur musicians and democratizes music creation.

However, this also raises the stakes for copyright law, monetization ethics, and the platform’s responsibility for protecting ownership. Some streaming platforms—such as SoundCloud—have already banned the monetization of fully AI-generated tracks, while others, like Spotify and YouTube Music, have launched impersonation reporting tools.

Deezer’s approach, however, balances openness with accountability. If AI-generated songs continue to surge, they could dominate most of the uploads on the platform within a year.

Platforms like Deezer are taking the first steps to differentiate between synthetic tracks from human ones to protect the essence of artistic originality.

So, are diffusion AI models remixing the music creation industry for good? How will music labels and platforms react to Deezer’s moves?

Let us know in the comments below!