Yandex Music: 37% of New Releases in H1 2026 Were AI-Generated

A Habr author who published the investigation "I found out that a third of Yandex Music is neural-network slop" a month and a half ago has presented a new algorithm for detecting AI tracks with near-100% accuracy, and has analyzed Yandex Music releases for the first half of 2026.

What the new analysis shows

In the first six months of 2026, 37% of all new releases on Yandex Music show signs of neural-network generation. Counting only tracks that earned likes from listeners, the share of AI music rises to 50%. In other words, every second popular release may have been created by a neural network rather than a human.

• The author's first article on neural-network slop on Yandex Music appeared a month and a half ago
• In the first half of 2026, 37% of the service's new releases contain signs of AI generation
• Among releases with likes, the share of AI music reaches 50%
• The new algorithm detects generated tracks with near-100% accuracy
• The algorithm does not require a GPU

How the picture changed in a month and a half

In the first article, the author estimated the share of neural-network slop on Yandex Music at roughly a third of the catalog. The new study, built on a more accurate algorithm and covering the entire body of releases from the first half of 2026, shows not merely a comparable figure of 37% but a more worrying signal: among tracks that listeners genuinely like and that collect likes, AI music already makes up half. That means neural-network tracks are not just filling the catalog as an afterthought; they compete with human artists for audience attention and land in playlists alongside them.

Why detection works without a GPU

Most modern AI-content recognition systems rely on neural-network models that need graphics cards for inference. The algorithm's author stresses that his approach works differently and does not use a GPU. That makes checking cheaper and more accessible for mass scanning of streaming catalogs, where releases number in the hundreds of thousands of tracks. Not needing expensive hardware is a key condition for applying this kind of detection at industrial scale, rather than only in a one-off study.

What this means for charts and royalties

If half of the releases with likes are generated by neural networks, the platform faces a question: how should it distribute audience attention and author royalties between human artists and generative projects? So far Yandex Music has not commented on the author's estimates and has not announced any tools of its own for labeling AI tracks in the catalog. This means listeners cannot yet tell neural-network music from human artists' tracks without a third-party detector.

What this means

If independent detection of AI music with near-100% accuracy really works without a GPU, streaming services and rights holders gain a cheap tool for labeling or filtering neural-network-generated content. The question is whether platforms like Yandex Music will actually use it.

Frequently asked questions

How much AI music is on Yandex Music in 2026?

According to the author's analysis, in the first half of 2026, 37% of all new releases on the service show signs of AI generation, and among releases with likes this share reaches 50%.

Is a GPU needed to detect AI music?

No. The algorithm's author stresses that the detector does not require a graphics card, unlike many other ML approaches to content recognition.