Scammers have found a reliable playbook on TikTok: take a real celebrity interview clip, use AI to swap in a fake voice and mouth movements, then run it as an ad promoting a fake rewards program. According to authentication company Copyleaks, this tactic is appearing regularly on TikTok, with Taylor Swift and Rihanna among the celebrities whose likenesses are being used without consent.
The ads follow a predictable pattern. They place the fabricated celebrity in a recognizable setting - a red carpet appearance, a podcast segment, a talk show clip - to borrow credibility from the real event. The AI manipulation replaces what the person actually said with promotional copy for shady services, most often fake rewards programs that collect personal information or payment details before delivering anything.
Deeply fakes - AI-generated videos that convincingly place real people's faces and voices into fabricated scenarios - have existed for years, but the production cost has collapsed. What once required significant technical skill and compute time can now be done with consumer tools in under an hour. That's changed the economics of scam advertising: high-quality fake celebrity endorsements are cheap to produce and hard for platforms to detect before they accumulate views.
TikTok has content policies that prohibit synthetic media used to deceive, but enforcement at the speed and volume of ad uploads is a real operational problem. Copyleaks, whose business is detecting AI-generated and plagiarized content, published these findings as part of ongoing work tracking AI misuse.
The scam format is not unique to TikTok - similar deepfake ad campaigns have surfaced on YouTube, Facebook, and Instagram. But TikTok's algorithm-driven discovery model means a fraudulent ad can reach a large audience in hours, well before human reviewers flag it. Until platforms build faster automated detection that specifically targets face-and-voice swap manipulation in ad inventory, this will keep working.