Quick answer: A late-April 2026 Copyleaks investigation found AI-generated deepfake ads on TikTok using Taylor Swift, Rihanna, Jennifer Aniston, Kim Kardashian, and Emma Watson to push a fake feature called TikTok Pay and a series of money-making schemes. None of those celebrities endorsed anything. The clips reuse real interview footage with AI-generated voiceover and redirect viewers to data-harvesting third-party sites.
Taylor Swift, on TikTok, is excited about a new feature called TikTok Pay. Rihanna explains that you can earn money just by watching content and giving your opinion. Jennifer Aniston tells you not to overthink it if the link opens for you.
None of those videos are real. None of those celebrities endorsed anything. The faces, the voices, and the scripts were all generated by AI, and TikTok delivered the videos as paid ads.
A late-April 2026 investigation by AI detection firm Copyleaks identified the campaign. The targeted celebrities also include Kim Kardashian and Emma Watson. The ads pushed a fictional feature called "TikTok Pay" and a series of money-making schemes that redirected viewers to third-party sites built to collect personal data.
What Copyleaks Found
Copyleaks documented dozens of paid ads on TikTok that used AI-generated likenesses of well-known public figures to promote scam services. The investigation surfaced two distinct ad clusters running through TikTok's paid-placement system.
The first cluster used celebrity faces. According to Copyleaks, a deepfake Rihanna tells viewers, "You literally just watch content and give your opinion." A deepfake Taylor Swift says, "if the page opens for you, don't overthink it." Each ad appears to take a brief clip of real interview footage and overlay AI-generated audio with a lip-sync adjustment, repurposing the celebrity's likeness for an endorsement they never recorded.
The second cluster, surfaced by Copyleaks and verified independently by Business Insider in March 2026, advertised AI apps that create sexualized deepfake images of real people, including features that digitally remove clothing from photos. The apps named in those ads included Soulove, Movely, POPGO, and Candy AI. The ads ran between December 2025 and February 2026 before TikTok removed them.
TikTok's policies prohibit ads that promote services for creating sexual content, and the platform requires realistic AI-generated content to be labeled. Both ad clusters violated those policies. TikTok removed the deepfake-app ads after Business Insider shared the list of videos. Copyleaks reports that the celebrity-likeness ads were similarly removed once flagged.
Why Paid Ads Are a Different Threat From User Posts
The story is not that scammers can make a deepfake of Taylor Swift. Anyone can do that now with public footage and a few minutes of compute. The story is that TikTok's paid-ad system delivered those deepfakes into users' feeds.
Paid placement breaks the usual defenses. A user who knows to scrutinize random viral videos assumes an ad has been vetted by the platform. The "Sponsored" label functions as a credibility signal in reverse: the viewer expects the platform took some responsibility for what appears under that tag.
It also breaks the discovery defense. A skeptical user can avoid sketchy accounts by curating who they follow. Nobody can curate around an ad. The platform decides which ads run and to whom they appear, which means the basic detection toolkit, the one built around vetting the account that posted a video, does not apply.
50+
sexually suggestive ads for deepfake-creation apps appeared on TikTok between December 2025 and February 2026, generating tens of thousands of views before removal.
Source: Copyleaks investigation, reported by Business Insider, April 2026.
How to Tell You Are Watching a Deepfake Ad
The Copyleaks ads are well-made by 2025 standards. They are not flawless. The visual and behavioral signals that work for any AI video work for these too, and a few show up specifically in celebrity-endorsement ad formats.
Lip sync on B, P, and M words. AI-generated lip movement struggles to fully close the lips on bilabial consonants. Watch a phrase like "make money" or "Bitcoin payment." If the lips do not seal between syllables, the audio was generated separately from the footage.
The off-platform redirect. Every fraud ad pushes you off TikTok. The legitimate version of any feature TikTok actually launches lives inside the TikTok app. If a celebrity in an ad tells you to "click the link" or "open the page," and the link goes to a domain you do not recognize, the entire premise is fraudulent.
The claim does not exist anywhere else. Real celebrity endorsements show up on the celebrity's verified accounts, in press coverage, and in the brand's official announcements. A claim that exists only inside a TikTok ad has not been made by anyone in a position to make it.
The face is right but the cadence is wrong. AI-dubbed audio often loses the speaker's natural rhythm. Pauses fall in unusual places. Emphasis lands on the wrong syllables. If a celebrity sounds slightly off in a way you cannot articulate, trust that signal.
The full visual checklist for TikTok specifically is at how to tell if a TikTok video is AI-generated. The behavioral signals matter at least as much as the visual ones.
What to Do If You See One
Do not click through. Do not enter any personal information on the destination page. Report the ad inside TikTok using the "report" option on the three-dot menu, and select "scam or fraud." Reporting is what flags an ad for human review.
If you already entered personal information on a site reached through one of these ads, change your passwords immediately and contact your bank if you submitted any financial details. The scam pattern is data harvesting first, monetization second, so the window between submission and consequence is often days or weeks. Full step-by-step reporting and evidence-preservation guidance is in what to do after you find a deepfake.
You can also paste the ad's TikTok URL below to check whether the Ledger community has already flagged it.
Think you found an AI video?
Paste the URL and let the Ledger community verify it. Free.
The Bigger Pattern
The shift to platform-served deepfake ads is the meaningful update here. Until now, most consumer-facing AI fraud rode on user-posted content. The viewer's defense was account vetting: short history, no original posts, weird follower-to-engagement ratios. Those signals do not apply to ads. The ad bypasses the account.
What still works is the check that has always worked. If a real public figure is endorsing something, that endorsement exists in more than one place. A celebrity's stated views appear on their verified accounts, in news coverage, and in official statements. If the claim only exists inside a video that was served to you by an algorithm, search for it somewhere else before you act on it.
Related Posts
- How to Tell If a TikTok Video Is AI-Generated: 7 Signs to Check Right Now: the canonical TikTok-specific detection guide
- That Celebrity Crypto Video Is Probably a Deepfake. Here Is How to Tell.: the user-posted version of the same fake-endorsement pattern
- Deepfake Doctors Are Selling Fake Supplements on TikTok: the same fake-authority playbook applied to medical claims
- The 6 Visual Tells That Instantly Give Away an AI Face on Video: the underlying detection skill that catches every variant of this scam
