Here’s how far-right extremists hide in TikTok’s earworms

Far-right extremists are exploiting TikTok’s “use-this-sound” feature as a Trojan horse for hate speech, with most of the offending videos staying online for months, according to new research published in arXiv, Cornell University’s preprint server.

Marloes Geboers of the University of Amsterdam and Marcus Bösch at Heinrich-Heine University in Düsseldorf scraped thousands of clips from German, British, and Dutch TikTok feeds. They found that over three-quarters of videos using extremist audio were still accessible four months after they were first captured.

Bösch says the project began when a familiar synth-pop loop transported him “straight back to the 1990s.” “Friends had a Nazi song on an actual tape,” he says. “Thirty years later it was on TikTok, and kids were filming their walk to school to that soundtrack.”

Tracking that song through TikTok led the researchers to dozens of trends in which seemingly harmless memes—such as users guessing what comes next in a song—masked what Bösch describes as “brutal, racist, misogynist and death fantasy lyrics.” The researchers then set up new, clean TikTok accounts trained to follow right-wing content in the three countries and began scrolling for extremist posts.

Their research uncovered extremist creators attaching hateful messages to everything from club classics by Gigi D’Agostino or Aqua to folk songs and even AI-generated tracks designed for niche audiences. “There’s Nazi techno, Nazi pop, Nazi folk—something for everyone,” Bösch explains. The goal was to lead users toward off-platform content intended to indoctrinate them into Nazi ideology. The team later checked the flagged videos two and four months after detection to see if they were still live.

TikTok’s moderation appears to struggle with audio-driven hate content. While the platform’s data indicates that text-based slurs are often removed promptly, 86% of videos featuring a racist song the authors call “Türke” remained online months later. Even overtly offensive material can evade detection: Bösch says he “stumbled upon a Hitler speech” reused in over 1,000 videos, often accompanied by visuals from Nazi propaganda. “You can’t argue that’s hard to see, hear or feel,” he says. “It shouldn’t be too hard to actually find these.”

In a statement to Fast Company, a TikTok spokesperson says the company employs a combination of technology and human moderation to detect and remove content that promotes hate speech or hateful ideologies, and says 94% of such content is taken down before it’s ever reported. “We continuously strengthen our enforcement by updating our detection tools, consulting experts, and partnering with local organizations,” the spokesperson adds.

Though Bösch acknowledges the challenges of moderating a platform at TikTok’s scale, he believes more can be done. “If a German court has banned a song, it shouldn’t be too hard to try and ban this song,” he says.