You put on a playlist to find something good to listen to. Checking whether the singer exists wasn’t supposed to be part of the process. But a song these days can fit your taste perfectly and still leave you wondering who, or what, made it.
This week, two music streaming services — Google’s YouTube Music and the French platform Deezer — announced features that could make that question easier to answer. YouTube Music is adding labels to AI-generated songs, while Deezer will let listeners filter fully AI-generated music out of search results.
For listeners who have been opening artist profiles and hunting for evidence of an actual musician, that could mean considerably less homework.
What’s changing when you press play?
YouTube says the new labels respond to something listeners have been asking for: "As these tools evolve, our listeners have told us they want transparency when they’re hearing an AI-generated track," the company wrote.
Starting Oct. 14, those labels will appear directly on YouTube Music’s player for songs created primarily using generative AI, including generated compositions and vocals. Tapping the label opens more information about the track. YouTube says it is working with music partners and using its own detection tools to identify AI-generated songs, with the labels also coming to the main YouTube app.
Deezer is offering an additional choice: leave those songs out of your search altogether. "We’ve seen a clear user demand for the ability to filter out AI-generated music," Deezer CEO Alexis Lanternier said in the company’s Oct. 14 announcement.
The toggle will launch worldwide before the end of 2026. Switch it on, and fully AI-generated tracks, albums, and related artists will disappear from search results. Switch it off, and they’ll be accessible again.
Deezer already keeps detected AI-generated songs out of algorithmic and editorial recommendations, so the toggle extends that control to music you actively search for. It targets fully generated music; songs with limited AI assistance within a broader human creative process are treated differently.
If you listen on Spotify or Apple Music, the approach looks a little different already. Spotify’s "AI Persona" badges, announced in Aug. with a rollout starting in mid-Sept., identify artist profiles depicting photorealistic AI-generated people. Spotify also said it would exclude those personas from editorial and algorithmic recommendations by default. The badge answers whether the artist’s public identity is synthetic, though it doesn’t specify how the music was made.
Apple Music’s reported "Transparency Tags," on the other hand get more specific about where AI was used, allowing labels and distributors to disclose generated content in recordings, compositions, artwork, and music videos. But that system relies on the companies delivering the music to declare it. An AI-generated album cover and an AI-generated vocal can receive different tags — provided someone reports them in the first place.
Your ears can only tell you so much
You don’t have to go looking for AI music to find it. Sometimes, your streaming service brings it to you.
That’s what happened to Mashable's Features Editor Rachel Thompson when Spotify’s Smart Shuffle slipped an AI-generated funk song into her queue. It fit alongside music she already liked, and she discovered how it was made only after investigating the artist. That’s a lot to ask of someone who just wants to listen to some funk!
But there’s a reason the song didn’t immediately stand out: in a Deezer/Ipsos blind test published in Nov. 2025, 97 percent of participants couldn’t correctly distinguish two AI-generated songs from a human-made one.
Once you start digging, though, even the usual signs of an established artist can leave you guessing. The Velvet Sundown attracted hundreds of thousands of Spotify listeners in 2025, complete with promotional images and song credits. Yet its supposed members had virtually no traceable existence beyond those suspiciously polished pictures, and the credits listed the band itself without naming a producer.
A label could put that information where listeners encounter the song, before they’ve opened an artist profile or searched for a band member. And plenty of people want that heads-up: in the same Deezer/Ipsos survey, 80 percent wanted fully AI-generated music labeled, while 45 percent of streaming users wanted the option to filter it out.
There’s money behind all those uploads
A song showing up in your queue is one thing. That song collecting royalties from thousands of fake listeners is another.
Deezer says it receives nearly 90,000 fully AI-generated tracks daily, but they account for just 1 to 3 percent of its streams. And even those plays aren’t necessarily coming from people: the company says up to 85 percent of streams of fully generated tracks were fraudulent in 2025.
North Carolina musician Michael Smith turned that gap between uploads and actual listeners into a multimillion-dollar scheme. On Oct. 6, he was sentenced to 18 months in prison after pleading guilty to conspiracy to commit wire fraud. Prosecutors said he generated hundreds of thousands of songs using AI, then had automated accounts stream them billions of times, collecting more than $8 million in fraudulent royalties.
The sheer number of songs helped him get away with it. By spreading the plays across thousands of tracks, Smith avoided drawing attention to any single one. Those streams pulled money from a shared royalty pool that, according to prosecutors, would otherwise have gone to musicians and songwriters with real audiences.
But "who gets paid?" gets complicated even when bots aren’t involved. A virtual artist’s face doesn’t tell you whose voice or writing you’re hearing.
Take FN Meka. In 2022, Industry Blackout criticized the virtual rapper for appropriating Black artists’ mannerisms and reproducing racist stereotypes, and Capitol Records subsequently dropped the project. Rapper Kyle the Hooligan also said he had supplied its voice without ever being paid. (Factory New co-founder Anthony Martini later clarified that humans had written and performed FN Meka’s vocals, walking back earlier claims about AI’s involvement.)
This Tweet is currently unavailable. It might be loading or has been removed.
That’s where a simple AI badge can run out of answers: It might flag the character or the recording without explaining who contributed what.
Still, YouTube Music’s labels and Deezer’s filter give listeners a place to start. A heads-up on the player could spare you a trip through an artist’s profile, while a search filter lets you decide whether you want fully generated music in the results at all.
How useful those controls become will depend on what platforms can reliably identify. If a song has no label, does that mean it’s human-made, or that it hasn’t been detected yet? And when human lyrics meet generated vocals, will the disclosure explain the mix?
For now, the promise is straightforward: finding something good to listen to could involve less detective work. The next step is making sure the information on your screen answers the questions that sent you searching in the first place.