Folk musician Murphy Campbell found songs on her Spotify profile that she never uploaded. They were AI-generated covers of her own performances, pulled from YouTube and distributed under her name without permission. The Verge AI reports on how this single incident spiraled into a full-blown nightmare involving fake profiles, copyright trolling, and the gaping holes in how streaming platforms protect artists.
Campbell discovered the fakes in January. Someone had taken her YouTube performances, run them through AI voice cloning, and pushed the results to streaming platforms. Two independent AI detectors flagged the tracks as probably AI-generated, confirming her suspicions.
Getting the songs removed wasn’t simple. Campbell says she “became a pest” before platforms acted. Even then, at least one fake track remains on Spotify under a duplicate “Murphy Campbell” profile. The real Murphy Campbell now shares her own name with an impersonator.
🎯 Then came the copyright claims
The day Rolling Stone published a piece about Campbell’s AI ordeal, things got worse. Someone using the name “Murphy Rider” uploaded videos through distributor Vydia and used YouTube’s Content ID system to claim ownership of Campbell’s material.
The kicker? The songs in question are all public domain. “In the Pines” dates back to the 1870s. Lead Belly recorded it. Nirvana covered it as “Where Did You Sleep Last Night.” Nobody owns it. Yet Campbell received a notice saying she’d now be “sharing revenues with the copyright owners.”
Vydia has since released the claims and banned the uploader. Spokesperson Roy LaManna told The Verge AI that only 0.02 percent of their 6 million-plus Content ID claims were found invalid. He also says Vydia has no connection to the AI covers uploaded to streaming services, calling the timing coincidental.
⚡ Why this matters
This case exposes three simultaneous failures in the music ecosystem:
- Distribution platforms don’t verify identity before publishing music under an artist’s name
- Streaming services lack adequate safeguards against AI-generated impersonation
- Content ID systems can be weaponized to claim public domain material
Spotify is testing a feature that would let artists manually approve songs before they appear on their profiles. Campbell isn’t holding her breath. “Every time an entity that’s that large makes a promise like that to musicians, it seems to just not be what they made it out to be,” she said.
What stands out here is how cheap and easy this attack vector is. You don’t need sophisticated tools. Grab a YouTube performance, clone the voice, upload through a distributor, and you’re collecting royalties under someone else’s name. Then file Content ID claims on public domain songs for extra revenue. The infrastructure that’s supposed to protect creators is being turned against them.
Campbell sees a systemic problem. “I think it goes way deeper than we think it does,” she says. The worlds of generative AI, music distribution, and copyright have created overlapping vulnerabilities that bad actors can exploit with minimal effort and almost no accountability.
Vydia says it received death threats over the incident, leading to office evacuations. Campbell acknowledges Vydia isn’t solely to blame but isn’t letting anyone off the hook either.
For independent artists, this is a warning. The tools to impersonate, distribute, and monetize someone else’s work exist today, and the platforms haven’t caught up. Full details are available in the original reporting from The Verge AI.