You have probably listened to an AI-generated song without realizing it. Maybe you shared it. Maybe your fans did. Meanwhile, someone else was monetizing an artist’s voice, style, or lyrics without their knowledge.
AI-generated music is happening at scale. Here are the four biggest ways artists’ voices and songs are being used right now:
- AI Covers: same song, different cloned voice
- AI Original Songs: brand new track, someone else’s stolen voice
- AI Changed Lyrics: real voice, words the artist never said
- New Music from Deceased Artists: new music from artists that can’t say no
1. AI Covers
We all know the appeal of a great cover song. But today, the internet is flooded with unauthorized AI covers. A quick search for #aicovers reveals the massive scale of the trend. Creators are even recreating songs with unauthorized cloned voices of major artists.
Ariana Grande fans mistook an AI version of “Kill Bill” for the real thing. A fake Taylor Swift “Fortnight” song fooled thousands. These songs generated millions of plays and real revenue for someone other than the artist.
The AI music tools making all of this possible are cheap, fast, and widely available. Platforms offer 500-plus artist voice models and can generate a cover in under 15 seconds for ten dollars a month. The barrier to entry is essentially zero. And the reason it keeps happening is just as simple: a traditional cover requires a mechanical license. An AI cover using a cloned voice does not fit that framework. The creator gets the credit and the revenue. The real artist gets nothing.
2. AI Original Songs
This trend is sneakier. A brand new song, new lyrics, new melody, generated in a real artist’s cloned voice.
In April 2023, “Heart on My Sleeve” went viral on TikTok and briefly topped both YouTube and Spotify. The producer had cloned Drake and The Weeknd’s voices using AI. Universal came in with legal threats and pulled it.
Today, social media accounts with hundreds of thousands of followers are being built entirely with AI original songs. AI music can now feature emotional inflection, breathing patterns, and vibrato.
One of the biggest trends we are seeing is religious content. There are thousands of videos where AI-cloned voices of major artists are made to sing Christian worship songs. Some of these videos have over 3 million views. The thumbnails and titles are designed to look like the real thing. “AI” is mentioned only occasionally, buried in the comments.
Every track made with an unauthorized AI-cloned voice claims two victims: the artist whose identity is exploited, and the label whose licensing revenue vanishes. Neither gets credit. Neither gets paid. Because these tracks are technically “original” works, standard copyright laws fail to protect them.
3. AI Changed Lyrics
Think Weird Al, but without permission. For decades, parody artists have rewritten popular songs for comedy, often with the blessing of the original artist. AI changes the equation. Now anyone can rewrite lyrics, clone a singer’s voice, and make it sound as though the artist recorded and endorsed words they never actually sang.
The harm extends far beyond lost revenue. An artist’s reputation, brand partnerships, and relationship with fans can all be affected when their voice is used with lyrics they never approved. Once the content begins circulating, corrections rarely travel as far as the original post.
From a legal perspective, these cases often present some of the strongest arguments for enforcement because they involve not only copyrighted material, but also unauthorized use of an artist’s name, voice, and likeness. The challenge is that those rights can only be enforced once someone discovers the content exists. By then, it may already have reached millions of listeners.
4. New Music from Deceased Artists
This one takes everything above and makes it more complicated, because the artist cannot speak for themselves. AI is being used to create new music from artists who have passed away, and nobody is there to say no.
The most high-profile example is Drake using AI-generated vocals of Tupac Shakur in his diss track “Taylor Made Freestyle” during his beef with Kendrick Lamar. The AI voice of Tupac, dead since 1996, was made to take Drake’s side in the feud. Tupac’s estate threatened to sue. Drake pulled the track! But by then it had already spread across social media and millions of people had heard it.
Done with permission, AI can be used in meaningful, creative ways. The Beatles used AI to complete John Lennon’s “Now and Then,” finishing a song he had started decades earlier, with the estate fully involved. But most of the time that is not what is happening. Estates are already dealing with unauthorized releases uploaded by fans who think they are paying tribute. The legal hook is the right of publicity, not copyright. That right passes to the estate and can be enforced just as it would be in life, but only if they know the content is out there.
The One Thing All Four Have in Common
In every scenario, the artist and the label are the last to know. By the time anyone notices, the content has already collected streams, built an audience, and in a lot of cases made real money for someone with zero right to it. Legal remedies only work when you know there is something to remedy. Catching it early is everything.
This Is What TraceID Is For
Vermillio built TraceID to monitor platforms so you know when your music or your artist’s identity is being used without permission. Not after the damage is done. Before it has a chance to spread.
And now, developers can use TraceID’s technology directly in their own applications. Vermillio just launched AI Guardrails-as-a-Service, a SDK (Software Development Kit) that lets AI music platforms, DSPs (Digital Service Providers), and developers bake artist rights and likeness preferences into their stack from day one. Labels define the rules once inside TraceID and those rules flow automatically to any platform using the SDK. No fragmented one-off deals. No guessing. Just a shared standard that puts artists and rights holders back in control.