音楽ストリーミングのDeezer、1日のアップロードの50%超がAI生成と発表Music streamer Deezer says more than 50% of daily uploads are AI-generated

Deezer, the France-based music streaming service, has said that more than half of the tracks uploaded to its platform each day are now generated by artificial intelligence, a figure that underscores how quickly generative tools are reshaping the economics and integrity of digital music. The disclosure matters because it suggests the volume of synthetic audio has come to rival, and in this case exceed, human output on a major service, sharpening questions about how platforms manage discovery, distribute royalties and protect the work of human artists.

The announcement continues a trajectory Deezer has been documenting for more than a year. The company previously reported that fully AI-generated tracks accounted for a smaller but rapidly rising share of daily uploads, and it has said the absolute number climbs into the tens of thousands of new files each day. The company frames the latest threshold, more than 50 percent, as evidence that the influx is not slowing. It is worth noting that a high share of uploads does not translate directly into a high share of listening; Deezer has indicated that AI-generated material draws only a small fraction of actual streams, meaning the flood is concentrated on the supply side rather than in what most users play.

To arrive at these numbers, Deezer relies on an in-house detection system it introduced to identify audio produced by popular generative platforms. The tool is designed to recognize signatures associated with services such as Suno and Udio, and Deezer has said it tags fully AI-generated tracks so listeners can see how a recording was made. The company has also said it excludes such content from algorithmic recommendations and editorial playlists, and it has moved to remove fully AI-generated tracks from royalty calculations in cases where it detects manipulation. Detection of this kind is imperfect and likely to remain an ongoing contest, as generation tools evolve and some uploaders attempt to obscure a track's origins.

The core concern for streaming services is not simply aesthetic. A large share of the industry's worry centers on fraud, where automated or low-quality tracks are uploaded at scale and paired with bot-driven streams to siphon royalties from a shared payout pool. Because most services distribute revenue on a pro-rata basis, every fraudulent or synthetic stream can dilute the earnings of legitimate artists. The sheer economics of generative tools, which can produce a finished song in seconds at negligible cost, make this a structural challenge rather than a temporary spike.

The surge is driven by a new generation of consumer-facing music tools. Suno and Udio can generate complete songs, including vocals and lyrics, from short text prompts, while other systems focus on voice cloning, mastering or instrumentation. These platforms have also become a legal flashpoint: the Recording Industry Association of America filed copyright suits against Suno and Udio, alleging their models were trained on copyrighted recordings without permission. The outcome of that litigation, and similar disputes globally, is likely to shape how freely such tools can operate and what obligations they carry.

Deezer's transparency also stands out against the broader industry backdrop. Larger rivals have taken varied approaches; Spotify has removed tens of thousands of tracks tied to spam and manipulation and introduced policies around AI voice impersonation, while YouTube and others have rolled out labeling and likeness-management features. Major labels, including Universal Music Group, have pressed for stronger protections and struck licensing conversations with AI developers, seeking to convert an adversarial dynamic into a paid, permission-based one. Regulators are watching as well, with disclosure and labeling rules for AI-generated content emerging in several jurisdictions.

For a mid-sized platform like Deezer, which competes against much larger incumbents, positioning itself as an advocate for transparency and human artists appears to be both a principled stance and a strategic one. The broader takeaway is that generative audio has shifted from a novelty to a dominant category of new uploads, at least by count. How the industry responds, through detection, labeling, revised royalty models and licensing deals, will likely determine whether AI music becomes a manageable layer of the catalog or a persistent strain on the systems built for human creativity.