AI-generated music is everywhere, but study shows only 3% of listeners can tell the difference

In 2025, music streaming service Deezer partnered with market research company Ipsos to conduct a study on people’s ability to identify AI-generated music. The result was disconcerting: only 3% of listeners could tell the difference between AI music and human-made music. This statistic is more alarming than ever, as Deezer reported in July 2026 that 90,000 AI-generated tracks were downloaded every day, representing over 50% of all new music downloads on the Deezer platform.

Currently, AI-generated music is completely legal. Additionally, it has proven to be very profitable for some people. An AI-generated rock band surpassed 500,000 listeners on Spotify, representing a minimum of $1,500 in revenue, even if each listener only streamed one song. One song in particular is “Dust on the Wind” by The Velvet Sundown, which has reached nearly five million streams on Spotify as of this writing. That’s at least $15,000 in revenue for a single AI-generated lead.

Some music platforms are taking action against AI music, but one of the uncomfortable truths about using Spotify (and other big-name services) is that AI garbage isn’t going anywhere anytime soon. These companies will benefit from listeners who serve AI-generated content without being able to tell the difference. If you want to avoid it, the best thing to do is to learn how to identify AI music yourself.

Learning to identify AI music is difficult. There are clear signs that a video is AI-generated, but it’s not so easy when your only tool is your ear. Here’s the trick, according to music lovers on Reddit: Listen for low-fidelity or “dull” sounds in the high frequency (treble) ranges of music. Audiophiles may also hear “lossy audio,” which is usually a sign of file compression, but it can also indicate poor quality AI generation. If you know how to read an audio spectrogram, you can sometimes detect these clues visually.

However, even with these techniques at your disposal, you certainly won’t be able to pick out AI music by ear 100% of the time. Forbes points out that the most important indices are not In music; they are around music. Be wary of artists who become popular suspiciously quickly or who have no online presence before their music is released. Learn how to read artist credits on your favorite music platform and tell if the people are real. From there, it’s a little easier to infer whether the music itself is real.

Deezer, the company behind the study on AI music detection, was also responsible for launching the first AI tagging system for music streaming. Deezer also excludes fully AI-generated tracks from algorithmic recommendations and refuses royalty payments to AI artists who commit fraud or stream manipulation on the platform.

Beatport went even further by completely banning AI-generated music from the platform. This follows Beatport’s partnership with Beatdapp, a music fraud detection company known for its advanced AI music detection tools. Other companies partnering with Beatdapp to combat AI-driven streaming fraud include Hangout FM, Audiomack and Napster.

But what if the big platforms are the only ones offering the music you’re looking for? Well, users agree that YouTube Music has a superior algorithm compared to other streaming services. This means YouTube is better at recommending music you’ll actually like, which can lead to the discovery of real artists that match your tastes. It’s not perfect, however, as some users report that YouTube still recommends AI-generated music from time to time and there is no option to exclude specific content from your taste profile. Until more companies understand why Gen Z hates AI, this is the best compromise most platforms will offer.