Her Own Voice, Someone Else's Claim: How YouTube's Copyright Machinery Turned on an Independent Singer

Updated:

In October 2026, a 50-second video disappeared from YouTube. It was an a cappella performance of "Bre Petrunko," a 19th-century Bulgarian folk song with no known author, sung and recorded by the musician and writer Samantha Shokin and viewed roughly 125,000 times over four years. The party that got it removed was not a major label or a publisher with a famous catalog. It was Media Nov Records, a company Shokin says she had never heard of, which by its own account runs about 500 YouTube channels and advertises automated takedowns and Content ID management.

A musician caught in a YouTube copyright claim loop (Photo: Kelly Sikkema)

The video came back about a day after a journalist started asking questions. That detail is the real story. YouTube's copyright system is built on the premise that the person holding the rights can reach the right lever — a dispute, an appeal, a counter-notice — and that the machinery will respond. In Shokin's account, every lever she pulled led back to the same closed door, and the one that finally worked was the press.

This piece reconstructs what is known about the case, explains how YouTube's copyright tools are supposed to work according to the company's own help pages, places the episode next to earlier cases of claims over material the claimant did not own, and weighs the counter-argument: that a system processing billions of claims a year is, by its own numbers, mostly uncontested.

The Timeline: A Folk Song, a Sample Pack, a Takedown

Most of what is known comes from 404 Media's reporting by Matthew Gault, published November 16, 2026, and a follow-up by Digital Music News two days later. Pieced together, the sequence runs like this:

  • During the pandemic: Shokin records a roughly 50-second solo vocal take of "Bre Petrunko" and posts it to SoundCloud, where it draws little attention.
  • About two years later: She uploads the recording to YouTube. Over the next four years it accumulates around 125,000 views.
  • Somewhere along the way: According to 404 Media, the vocal ends up, without her knowledge, in a commercial sample pack — Loopmasters' "Folk Legends" — and from there in dozens of DJ and electronic tracks, one of which has more than 4 million views. 404 Media reports the distributor apologized and paid her royalties on the pack.
  • October 2026: YouTube removes her original video after a copyright complaint from Media Nov Records.
  • After the removal: Shokin files two counter-notifications. YouTube rejects both. Its final reply, as quoted by 404 Media, said it was "unclear to us whether you have a valid reason for filing a counter notification," and told her to contact the copyright holder. Shokin told 404 Media she could not get a human being on the line.
  • After 404 Media contacts Media Nov: The company first says earlier emails went to spam, then asks Shokin for original audio files, recording evidence and publication dates. Minutes after she provides them, it withdraws the complaint. YouTube restores the video about 24 hours later; a YouTube spokesperson confirmed the reinstatement to 404 Media.

Two points in this timeline deserve a flag. First, the coverage does not pin down exact dates for the counter-notices or the withdrawal. Second, the two outlets describe the downstream reuse differently: Digital Music News's standfirst refers to "DJ and AI-generated remixes," while 404 Media's account of how her voice spread centers on the sample pack and human-made DJ tracks, and attributes the AI-generated material to Media Nov's own channels. This article follows 404 Media's more detailed version.

Who Filed the Claim

Headlines have called Media Nov an "AI company." The reporting supports a narrower description. According to 404 Media, Media Nov presents itself as a distribution and rights-management firm; its website emphasizes automated DMCA takedowns, YouTube Content ID management and "AI content recognition," and claims a history dating back to 1996, although the Wayback Machine shows little trace of it before last year. Irish corporate filings reviewed by 404 Media show it was incorporated a little over a year ago at the Black Church in Dublin, a converted church now used as offices, with two shareholders listing addresses in Turkey. A second listed address is an apartment building in Marina del Rey, Los Angeles.

The company told 404 Media it runs 500 channels. Reporters found four, each with hundreds of videos: looping lo-fi-style beats paired with AI-generated images of sports cars and anime-style women, often drawing fewer than 100 views, plus "random acts of kindness" Shorts, some carrying a Douyin watermark. So the AI element is real, but it is on the output side — mass-produced channel content and automated enforcement — rather than an AI model that generated a copy of Shokin's song.

What remains unknown is the most important part: what reference Media Nov claimed to own, and why it matched Shokin's recording. None of the coverage says. One plausible route — and this is the author's inference, not a reported finding — is the sample pack. Once a vocal circulates in a commercial loop library, it can end up inside many derivative tracks; if any of those tracks is registered with a rights-management service as a reference, the original singer's own upload can look like the copy. Media Nov's own explanation pointed elsewhere: it raised Philip Koutev's 1963 arrangement of the song and argued that arrangement rights must be distinguished from rights in a particular sound recording. 404 Media notes that Koutev died in 1982 and that the rights to his recording are held by the Smithsonian, which does not appear to have been involved. Shokin's response, quoted in both outlets, was that the dispute was about "the voice and performance," not the arrangement.

How the Machinery Is Supposed to Work

YouTube runs two distinct copyright tracks, and the difference matters for this case.

Content ID is the automated matching system. According to YouTube's help page on who qualifies for Content ID, partners must "have the exclusive rights to the material that's evaluated," must provide evidence of the content for which they control exclusive rights, and sign an agreement that only exclusively owned content may be used as a reference. The same page lists material that often does not qualify: mashups, compilations, remixes, unlicensed music and material licensed without exclusivity. YouTube says it can disable reference files, require manual review, switch off Content ID or terminate a partnership when it finds abuse or error (Content eligible for Content ID).

When a Content ID claim lands, the dispute process gives the claimant 30 days to respond to a dispute; if it does nothing, the claim expires. If the claimant reinstates the claim, the uploader can appeal, and the claimant then has 7 days. To keep a claim alive after an appeal, the claimant must file a formal copyright removal request — a legal process that, if valid, takes the video down and puts a strike on the channel.

Copyright removal requests (takedowns under the US DMCA) are the second track, and they are what 404 Media describes in Shokin's case: a removal, not merely a claim that redirects ad revenue. The coverage does not establish whether the complaint began as a Content ID match. The remedy for a takedown is a counter notification. Once a valid one is forwarded, the claimant has "10 US business days" to show it has taken legal action; otherwise the video is restored. That clock comes from the statute itself: 17 U.S.C. § 512(g) requires a service provider to restore material "not less than 10, nor more than 14, business days" after receiving a counter notice, unless the complainant reports that it has sued. Section 512(f) makes anyone who "knowingly materially misrepresents" that material is infringing liable for damages, "including costs and attorneys' fees."

Set against this design, Shokin's case shows where the loop formed. Her counter-notices never started the 10-day clock: YouTube rejected them before forwarding, telling her it was unclear whether she had a valid reason. The claimant, meanwhile, was not answering email. The safeguard that is supposed to put the burden on the claimant — sue or let the video go back up — was never triggered. YouTube has not publicly explained why the counter-notices were rejected, and as of publication no further comment from YouTube beyond confirming the reinstatement could be found.

Not the First Time

Claims over material the claimant did not create are an old problem on YouTube; what changes is the scale and the business model behind them.

Ten hours of static (2018). The Australian music technologist Sebastian Tomczak uploaded a 10-hour white-noise video in 2015. By January 2018 it had collected five Content ID claims from four different rights holders, mostly publishers of sleep-aid noise tracks, who chose to monetize his video rather than take it down. After disputes and press coverage, the claims were released (Tubefilter; EFF). The pattern is familiar: a generic or unownable sound, an automated match, and a fix that arrived only with publicity.

MediaMuv (2021–2023). The largest known abuse of YouTube's monetization machinery was deliberate. According to the IRS Criminal Investigation release on the case, Jose Teran and Webster Batista formed MediaMuv L.L.C. "for the purpose of stealing royalty proceeds," falsely claimed ownership of more than 50,000 song titles and received about $23.4 million. Teran was sentenced to 70 months in June 2023; Batista pleaded guilty to conspiracy and wire fraud in 2022 (The Register). There is no reported connection between MediaMuv and Media Nov, and nothing in the reporting suggests Media Nov acted fraudulently; the case is relevant because it shows that a company sitting between rights holders and YouTube's tools can claim catalogs it does not own for years before anyone checks.

Shokin's case sits between those two poles. It is not a pure algorithmic accident, because a company filed a removal request. And there is no evidence it was a fraud scheme. It is the middle ground Steiner described to 404 Media: an operation that is probably understaffed, leans on automated tools and may be "taking down far more than they have the right to."

The Case for the System — and Its Blind Spots

It would be easy to read this story as proof that YouTube's copyright system is broken. YouTube's own numbers make a more complicated case.

According to YouTube's copyright transparency data for 2025, as summarized by GIGAZINE, the platform processed about 2.5 billion Content ID claims that year, 99.76% of them generated automatically. Only 0.52% of automated claims were disputed, compared with 1.43% of manual claims — a gap that suggests human-filed claims are more likely to be contested. For removal requests submitted through the public webform, 6.44% were flagged as abusive and 6.65% as invalid. The same summary cites more than ¥2 trillion in ad revenue paid to rights holders through Content ID claims between 2021 and January 2025. For the large majority of musicians, labels and publishers, Content ID is the only practical way to find and monetize unauthorized uploads at YouTube's scale; without it, the alternative is manual takedown notices, which are slower and harsher.

When disputes do happen, uploaders often win. YouTube's first transparency report, covering January to July 2021, counted 722.6 million claims, of which about 3.7 million were disputed and roughly 2.2 million — 60% — were resolved in the uploader's favor (Tubefilter).

These figures have limits. "Resolved in favor of the uploader" includes claims that simply expired after 30 days without a response, so it is not a count of confirmed errors. A low dispute rate may reflect deterrence as much as accuracy: disputing means risking a strike, and counter-notices require a legal name, address and phone number that are passed to the claimant. And the published statistics do not separate out the scenario in Shokin's case — a counter-notice rejected before it reached the claimant.

The account of this case has limits of its own. It rests largely on one party's description, relayed through two outlets. Media Nov disputes the substance: it says the withdrawal "does not constitute a waiver of any rights" and that it intends to take legal action. Its stated reason for withdrawing — that the video had earned no revenue or views — conflicts with the roughly 125,000 views reported by 404 Media, but the company has not published evidence for its arrangement-rights theory either. Until a court or YouTube weighs in, the fairest summary is that Shokin says the claim was baseless, Media Nov says it has rights it has not yet detailed, and the claim was withdrawn once she produced proof of her recording.

How This Fits a Bigger Pattern

Two earlier pieces on this site looked at adjacent failures. False Positives from AI Detection Tools Threaten the Indie Musicians They're Supposed to Protect examined detectors that turn a probability score into a verdict; Shokin's case shows the same problem in copyright enforcement, where a match or an assertion becomes a takedown before anyone checks it. DistroKid has been quietly taking down songs in response to UMG lawsuit traced how a distributor under legal pressure pulled tracks with little explanation; here the pressure point is the other end of the pipeline — a rights-management intermediary filing claims inside YouTube itself.

The common thread is cost. Automated enforcement is cheap to file and expensive to contest, and the expense falls on whoever has the least time and money. As Steiner put it to 404 Media, if everyone you wrongly target is small and you are hard to pin down, "it can be difficult to get a remedy."

What Ended the Loop — and What Should

What finally resolved Shokin's case was not a policy but a set of files: original audio, recording evidence and publication dates. That is practical advice for any independent musician who records traditional or public-domain material — keep session files and dated originals, because they are the proof a claimant or a platform will ask for. But it should not take a journalist to get someone to request them.

The fix implied by YouTube's own rules is modest. A counter-notice that meets the legal requirements is supposed to put the burden on the claimant, not be screened out at the door. Rights-management intermediaries that use Content ID or file takedowns at scale should be easy to reach, and YouTube's existing power to disable references or terminate partners is only a deterrent if it is visibly used. Shokin's verdict, quoted in both reports, was blunt: "It feels like a broken system." The data suggests the system works most of the time. Her case shows what happens to the people it fails, and that the public figures do not count them.

Sources