Deezer hits 50% AI uploads, then pledges to delete fraud-suspect tracks

On Tuesday (July 21), Deezer announced that, for the first time, fully AI recordings made up more than 50% of new music uploads on its service at peak level in June 2026, according to a press release. The company says that translates to 90,000 AI tracks. That is not a gentle trendline. It is a sudden shift in what “new music” means on a major platform, and it forces every executive question you can’t dodge: what do you do when AI content stops being the tail and becomes the feed?

Deezer’s answer is twofold, and it targets both trust and catalog bloat. First, it says it will remove any AI track it suspects has been used in streaming fraud scams. Second, it will remove AI tracks that have not been streamed for six months or more. The policy matters because Deezer is already deep in the weeds on this issue. In an earlier update, it reported that up to 85% of streams on fully AI tracks were fraudulent, and when streaming manipulation is found, Deezer excludes those streams from royalty payments.

To understand why this is such a big deal, you have to zoom out to the structure of the music economy. Streaming platforms are paid attention and royalties, and both depend on the integrity of streaming numbers. When AI-generated tracks are used to inflate plays, you get payment dilution, disputed claims, and a growing headache for rights holders. Deezer’s approach is essentially triage plus accountability: identify fully AI recordings, detect suspicious behavior, and then stop the system from paying out when the numbers cannot be trusted.

This is not Deezer improvising. The company says it has been tracking the influx of AI-generated content since January 2025, when it first announced plans to track and tag fully AI-generated recordings. At that time, its proprietary AI detection system flagged 10% of daily uploads as AI-generated. Deezer then pledged that detected songs would be removed from algorithmic recommendations and would not be eligible for editorial playlisting, but would still remain on the site for fans who wanted to stream them.

Now the “remain on the site” part is getting squeezed. The new vow goes beyond recommendation suppression into actual catalog removal, and it adds a time-based filter: AI tracks with no streams for six months or more are out. That is a direct response to a problem that is both financial and operational. In a previous interview with Billboard, Aurelien Herault, chief innovation officer at Deezer, framed storage as a serious issue. He said Deezer had already removed some content because it uses too much space, adding that it is not only an economic issue but also an ecological one. The implication for leadership teams is straightforward: if AI uploads keep climbing, “we’ll just host it” becomes a cost and sustainability problem, not just a content policy.

The distinction Deezer draws between “fully” and “partially” AI-generated songs is also crucial, because it limits where the detection system draws its line. Deezer’s AI detection tool, like nearly all others on the market, flags 100% AI-generated recordings from prolific models like Suno and Udio. It does not flag or penalize songs partially generated by AI. That means Deezer’s deletion and fraud-suspect policies apply to a specific bucket of recordings, not every track that might have an AI ingredient. For boards and compliance leaders, that distinction affects enforcement risk. Policies that are too broad can punish legitimate releases. Policies that are too narrow can leave gaps that fraudsters exploit.

Deezer also gave numbers beyond the June 2026 “more than 50%” headline. The company says it detected and tagged more than 13.4 million fully AI-generated tracks on the service in 2025. And it positions itself as a long-running fighter against fraud and payment dilution related to AI music, claiming it has been doing so for almost two years. Alexis Lanternier, CEO of Deezer, tied the update to a broader industry problem: if half of daily uploads are AI-generated, the platform is taking additional steps to safeguard the rights of artists and songwriters while maintaining focus on music that fans actually love. He also said there is a pivotal moment and that if the industry aligns on shared standards, it can build a healthy and sustainable music ecosystem where great music wins.

For executives at platforms, labels, and rights-collecting organizations, the strategic stakes are immediate. Deezer is effectively turning “AI content management” into “AI risk management,” with concrete rules for deletion when tracks look fraudulent or stale. If other services follow, it could reshape how new music is surfaced, how fraud detection systems are audited, and how royalties are defended. If they do not, platforms that refuse to act may end up paying for it indirectly through higher fraud rates, more disputes, and reputational damage when artists and songwriters feel the dilution.

The real headline underneath the headline is this: Deezer is no longer just reacting to AI music. It is redesigning the rules of the feed while the feed itself is changing faster than anyone was planning for. Half of new uploads being fully AI recordings changes the math for content moderation, detection accuracy, storage strategy, and payments. And once you cross that line, the only viable next step is operational, not philosophical. Deezer says it has the technology to keep track of development, and it wants shared standards. Today, it is starting with deletion.