An AI content mill used copyright automation to erase the musician it copied
Musician Samantha Shokin recorded a 50-second performance of the public-domain Bulgarian folk song “Bre Petrunko” during the pandemic. It later accumulated 125,000 YouTube views and was sampled by electronic producers. Last month, a company called Media Nov Records claimed the recording and YouTube removed Shokin’s original, 404 Media reports.
Shokin filed two counter-notices with original files and publication history. YouTube rejected both and directed her to the claimant. Media Nov did not respond to her or the sample distributor. Its site advertises automated takedowns and AI content recognition; corporate records indicate it was founded a little over a year ago despite a claim that it has existed since 1996. The company told 404 Media that it operates 500 channels, many filled with low-view, AI-generated music and imagery.
After the reporter contacted Media Nov, it asked Shokin to prove that she owned her own recording. It later withdrew the takedown while reserving supposed rights, and YouTube restored the video. Media Nov never identified which work in its catalogue had triggered the claim.
The system’s asymmetry is the story. A claimant can obtain enforcement through automation; the creator must assemble evidence, locate an opaque company and reach a human at the platform. AI makes producing hundreds of channels cheap, while Content ID turns that volume into claims with immediate practical force. A workable rights system would require the claimant to show a matching protected recording before removal, especially when the target supplies earlier source material.