A teenager scrolling TikTok late at night sees a familiar face. It is Taylor Swift, looking straight into the camera, telling them about a giveaway. Free cookware. A bonus card. A limited offer if you click right now. The voice is hers. The face moves the way her face moves. And it is completely fake.
This is not a one-off prank. Over the past year, deepfake ads using the likenesses of Taylor Swift, Rihanna, Kim Kardashian, and other A-list celebrities have flooded TikTok and other short-video platforms, funneling thousands of people toward scams that harvest personal data and credit card numbers. The technology is now cheap enough, fast enough, and convincing enough that a stranger with a laptop can put words in a global superstar's mouth in an afternoon. For anyone who teaches, parents, or works with young people who live inside these apps, this is worth understanding clearly.
What the scam actually looks like
The pattern is remarkably consistent. A short video shows a celebrity appearing to endorse a giveaway or a too-good-to-be-true deal. Swift "announces" a Le Creuset cookware giveaway. Rihanna "promotes" a product drop. The clip looks like it was lifted from an interview or a livestream, but the audio has been swapped and the lips have been re-synced to match new words the celebrity never said.
The video is the bait. The hook comes next. Viewers are told to click a link, answer a few quick survey questions, and pay a tiny shipping fee to claim their prize. That small fee is the whole point. To pay it, you hand over your name, address, and full card details on a page that has nothing to do with the celebrity or the brand. Sometimes the charge is a modest one-time hit. More often it quietly enrolls the victim in a recurring subscription buried in fine print, or the card data simply gets resold. The "free gift" never arrives because it never existed.
What makes these especially effective is that they often run as paid ads, not just organic posts. Paid placement means the platform's own ad system pushes them into feeds, lending a veneer of legitimacy and reaching far more people than a random account ever could.
Why celebrities, and why now
Scammers borrow famous faces for the same reason advertisers pay millions for endorsements: trust transfers. If you have followed someone for years and feel like you know them, your guard drops when they seem to recommend something. Swift in particular is a useful case study because her fanbase is enormous, intensely loyal, and skews young, which is exactly the audience scammers want and exactly the audience least likely to assume a video could be fabricated.
The "why now" part is simpler. Until recently, a convincing face-and-voice swap took real skill, real time, and real money. Generative AI collapsed all three. Voice cloning tools can reproduce a recognizable voice from a short sample. Video models can re-animate a face to match new speech. The barrier to entry has fallen so far that the limiting factor is no longer technical ability but willingness to defraud people. That is a bad trade for the rest of us.
It is also why Swift's team has reportedly explored trademarking aspects of her likeness. When your face becomes a tool other people can pick up and use to lie, the legal system built for an analog world starts to look thin.
How to spot one
The good news is that today's deepfakes still leave fingerprints if you slow down and look. A few things to watch for:
- The mouth is slightly off. Re-synced lips often do not quite line up with the sounds, especially on hard consonants. Watch the mouth, not the eyes.
- The audio feels flat or oddly smooth. Cloned voices can lack the small breaths, stumbles, and shifts in tone that real speech has. If it sounds like a polished read with no texture, be suspicious.
- The offer creates urgency. "Only today," "first 500 people," "claim now." Pressure to act fast is the oldest trick in fraud, AI or not.
- The link does not match the brand. Real giveaways from real companies live on the company's own verified site. A misspelled domain or a random link in a TikTok bio is a red flag.
- A celebrity is asking for your money or data directly. Stars do not run cookware giveaways that require your credit card for shipping. Full stop.
- It only exists in the ad. If a megastar were truly giving away thousands of products, it would be everywhere, not just in one suspiciously targeted video. A quick search for the supposed promotion usually turns up nothing, or turns up warnings.
None of these is foolproof on its own, and the tells will keep shrinking as the tools improve. The durable defense is not visual but behavioral: treat any unsolicited offer that wants your money or your data as guilty until proven innocent, no matter whose face is attached.
Why this matters in a classroom
You might wonder why a plagiarism and AI-detection company cares about TikTok scam ads. Because the underlying skill is the same one we want students to build everywhere: the habit of asking whether what is in front of you is real.
Students are growing up in an environment where a video is no longer proof that something happened, where a familiar voice is no longer proof that a person said something, and where the line between authentic and synthetic is genuinely hard to see. That is the same world in which they will be asked to evaluate sources for an essay, decide whether a quote is trustworthy, and judge whether a piece of writing was produced by a human or a machine. Deepfake scams are media literacy with stakes. A student who learns to pause and verify before clicking "claim now" is practicing exactly the reasoning they need to be a careful thinker, not just a careful shopper.
It is also a chance to retire a comforting myth. Many adults still assume young people are naturally savvy about technology and therefore immune to this stuff. They are fluent, but fluency is not the same as skepticism. Being fast on an app does not make you good at detecting deception on it. Anyone can be fooled by a good fake, and the people most confident that they cannot be are often the easiest marks.
What to actually do about it
For parents and teachers, the move is not to ban the apps or panic about AI. It is to talk plainly and often. Show a known deepfake clip and walk through the tells together. Make "I'll check first" a normal reflex rather than a sign of paranoia. Set a simple household or classroom rule: no clicking offers, no entering card details, no "small shipping fees," without an adult or a second opinion first. Report the scams when you see them, because platforms do respond to volume, and reporting is one of the few levers ordinary users actually have.
And reset expectations honestly. The fakes will get better, not worse. The face you trust is no longer evidence. The voice you recognize is no longer proof. What still works is the boring, durable stuff: slow down, check the source, never let urgency make the decision for you.
If a celebrity in your feed is asking for your credit card, the safest assumption is the simplest one. It is not them.