Deezer finds 97% of listeners can't spot AI-generated tracks

Even if you consider yourself an audiophile, generative music can fool you without breaking a sweat.

Music streaming service Deezer teamed up with research firm Ipsos for an eye-opening experiment. They played three songs to 9,000 adults across eight countries—including the US, UK, Canada, and Japan—and asked them to identify which tracks were completely generated by artificial intelligence.

Nearly all respondents (97%) failed to tell the difference.

Fooled ears, uneasy listeners

Getting tricked by an algorithm did not sit well with most listeners. Among those who could not tell human from machine, 71% said they were surprised, and 52% admitted they felt uncomfortable about being unable to distinguish the two.

Listeners are caught in a strange paradox. Around two-thirds showed curiosity and said they would give AI tracks a spin at least once. Yet four out of five respondents (80%) insisted that AI music must be clearly labeled by streaming platforms.

Furthermore, 45% said they would filter out AI-generated music entirely if their streaming service offered the option, while 70% believe machine-made songs threaten the livelihoods of real artists. Over half also worry the technology will flood platforms with generic, low-quality filler.

The flood is already here

Deezer has good reason to push this conversation into the spotlight. The streaming platform revealed it now receives roughly 50,000 AI-generated tracks every single day, which pushed it to launch an AI detection tool back in January.

This is not a hypothetical future threat. Just recently, an AI-powered country act named Breaking Rust actually topped Billboard’s country digital song sales chart.

Streaming giant Spotify is also moving fast. Last month, it signed landmark deals with Sony, Universal, and Warner to build licensed AI music products.

Hardware makers are speaking up too. Paul McCabe, senior VP at gear manufacturer Roland, noted that while AI can amplify human creativity, his company opposes consumption platforms that push purely AI-generated tunes to compete with human creators. Roland is currently working with partners on tools to trace musical provenance and verify whether tracks are synthetic.

Why it matters

For everyday listeners, music is not just pleasing frequencies; it is fundamentally about human connection and lived experience. Now that algorithmic generation can effortlessly mimic standard production values, technical fidelity is no longer the bottleneck—transparency is.

For the music industry, receiving tens of thousands of synthetic tracks daily amounts to automated spam. If streaming services fail to separate human craft from machine-generated wallpaper, payout pools will dilute, and human musicians could be starved out of the discovery feed.

Deezer’s move here is shrewd. By quantifying consumer anxiety, it is framing its proprietary AI detection engine as essential infrastructure while putting pressure on rivals. Meanwhile, Spotify’s pact with the major labels looks like an attempt to domesticate the beast before it destroys their catalog value. In our view, platforms that drag their feet on clear labeling risk alienating their paying subscriber base, turning their music apps into sonic junk drawers.