After a fake Drake song went viral, the music industry is now building tools to track and label AI-generated music

Drake

The music industry is still dealing with the aftershocks of a fake Drake song that went viral last year and now, it’s racing to keep up with AI-generated music.

The song, Heart on My Sleeve, which sounded like a duet between Drake and The Weeknd, fooled millions. But the artists had nothing to do with it. It was created using artificial intelligence, and its success exposed just how quickly AI can blur the lines between real and fake.

Now, instead of trying to stop AI-generated music completely, the focus is shifting to tracking and tagging it, according to a report by The Verge. A new system of tools and technology is being built to detect AI-created songs early — even before they’re uploaded or streamed.

Platforms like YouTube, SoundCloud, and Deezer have already started using tools that scan tracks when they’re uploaded. These tools flag fully or partially AI-generated songs and influence how they show up in search or recommendations. Some companies are going even further, building tech that can track AI’s role in a song from the training data stage to the final release.

One such company, Vermillio, has developed a system called TraceID that breaks a song down into parts — like lyrics, melody, or vocal style and identifies which sections were created by AI. This helps labels and platforms decide if a song needs to be licensed before it goes public. Vermillio estimates that AI music licensing could grow from $75 million in 2023 to $10 billion by 2025.

Meanwhile, Musical AI is working on systems that can trace a song’s origin from the very beginning when an AI model starts learning instead of just scanning the final product. The goal is to measure “creative influence,” not just spot obvious copies.

There’s also a push to stop AI from being trained on artists' work without permission. Tools like the Do Not Train Protocol allow musicians to label their content as off-limits, though experts say it needs to be handled by a neutral, trustworthy body to truly work.

Deezer, for example, says it already detects about 20% of its daily uploads as fully AI-generated. These songs aren’t removed but aren’t promoted either. The platform plans to start labeling AI content for listeners soon.