If you’ve spent any time on TikTok recently, you might have felt a strange sense of déjà vu. You’re scrolling through your For You Page, watching a creator talk about their daily routine or recount a funny story, but something about their face just feels… off.
It’s not your imagination. A sophisticated new trend is taking over social video feeds: AI video hijacking.
Unlike early deepfakes or obvious synthetic AI avatars, these videos don’t start from scratch. Fraudulent accounts, often operating as automated engagement farms, are downloading real creators’ videos, isolating their original voice audio, and reskinning the footage using AI face-swaps.
It’s video identity theft and digital puppetry rolled into one. While the technology is getting scarily realistic, the seam marks are still there if you know where to look.
Here are 5 dead giveaways that the TikTok you’re watching has been hijacked by AI.
1. The lip-sync detachment
Because the AI is trying to map a new synthetic face onto an existing audio track that was scrubbed and isolated from the original person, the audio and visual timing almost always fall out of sync. Pay close attention to plosive sounds (words starting with P, B, D, or T).
In a real video, a person’s lips press together firmly to make these sounds. In a reskinned AI video, the lips will often float open, misalign, or look slightly detached from the actual voice's cadence.
@nia.noir31 ♬ original sound - Nia Noir
2. Lighting disconnect (face vs. body)
Pay attention to how the light hits their skin. Real-world environments have shifting, unpredictable light, whether it’s sunlight through a window, streetlights on a drive, or passing shadows. But generative AI overlays struggle to adapt to these changes in real time.
Watch the creator’s forehead and cheekbones as they move: if the light on their body shifts naturally, but their face stays frozen in the exact same lighting, you’re looking at a digital mask.
@aitheftexposed ♬ original sound - Abigail liddle
3. Blurring at the edges
AI video tools have gotten scarily good at blending synthetic faces onto real bodies, but seamless edges are still their Achilles' heel. Where does the face end and the original person begin? That border is where the software regularly fails.
Look closely at key contact points like the neck, collarbone, and hairline. You might spot weird blurring, pixel shifting where the jaw meets a shirt collar, or hair strands that vanish unnaturally into forehead skin.
The biggest giveaway happens when hands come into frame: if the creator reaches up to adjust their hair or touch their chin, the face overlay will often momentarily distort around their fingers.
@lilamonroetts ♬ Storytelling - Adriel
4. Pupil drift and the uncanny gaze
Human eyes continuously micro-adjust to focus on a camera lens. When an AI model maps a generated face onto a real person's eye line, the pupils often struggle to lock onto the viewer.
This results in the uncanny gaze: eyes that seem to float, swim around the socket, or look slightly past the camera even when the head is facing forward. If the person looks like they’re staring at a ghost just behind your shoulder, unblinking, trust your gut.
@aitheftexposed ♬ original sound - Ai Theft Exposed 🧾
5. Audio phasing and reverb drops
To make the stolen video work, the account owners rip the audio straight from a real creator’s existing post. But to separate that original voice from its background music or room noise, they have to run it through AI isolation software, and that cleanup process leaves distinct digital scars.
Put in headphones and listen carefully. Even though the dialogue itself sounds completely natural, because it is a real person talking, the isolation software can mangle audio quality.
Listen for a robotic, muffled tone, or dead silence between words where natural room echo, breathing, or background hiss should be.
If the video looks high quality, but the voice sounds tinny, muffled, or isolated in a silent background, you’re likely listening to ripped audio pasted onto a synthetic face.
@lilamonroetts ♬ original sound - Lila Monroe
Why this is happening, and why it’s so alarming
Beyond the immediate nuisance of TikTok Shop spam and account farming, the real issue is that creators lose all ownership of their image.
When an account steals your face, you can end up "endorsing" products you've never used, saying things you'd never agree with, or unwittingly scamming your own audience, with almost no easy way to get the content taken down quickly.
Have you stumbled across any obvious AI fakes on your feed lately? Drop a link or let me know what tipped you off in the comments.
