Skip to content
Toolncy
All articles

How to spot a deepfake candidate in a video interview

Face-swap tools are now good enough to fool a tired interviewer. Here are the tells that still give them away, and what to do when you suspect one.

Toolncy TeamUpdated 6 min read

Short answer: watch the edges of the face, the sync between lips and voice, and how the image behaves when the candidate moves quickly or touches their face. Then confirm identity across rounds instead of trusting any single call.

Remote hiring removed the handshake from the process, and with it the simplest identity check. Real-time face-swap and voice-cloning tools are now cheap and easy to run on a gaming laptop. Analysts expect the problem to grow: Gartner forecasts that by 2028, one in four candidate profiles worldwide could be fake.

Visual tells that still work

  • Edges and occlusion. Ask the candidate to adjust their glasses, wave a hand across their face or turn to the side. Real-time swaps often smear, flicker or briefly show the original face when something passes in front of it.
  • Lighting that doesn't match. The face may be lit differently from the neck, hair or room. Look for a face that stays evenly lit when the candidate moves.
  • Lip sync. Plosive sounds such as "p", "b" and "m" need closed lips. Many swaps lag here by a few frames.
  • Blink and micro-expressions. Reactions that are too even or too slow, especially when the candidate laughs or is surprised, are worth noting.

Process tells

  • The camera is "broken" for the first round but works perfectly later.
  • The candidate refuses any small request to move, turn or change lighting.
  • Their voice, accent or knowledge changes noticeably between rounds.

What to do when you suspect a deepfake

Don't accuse anyone on the call. Note the timestamps and what you saw, then ask for a short, structured identity step: a live ID check, or a final round in person or on a fresh link. Honest candidates rarely mind. Fraudulent ones usually disappear.

Treat every signal as a reason to verify, never as a verdict on its own.

Where tooling helps

People get tired, and synthetic video keeps improving. InterviewShield checks every frame for blending artefacts and lip-sync drift, and compares the face across interview rounds. It hands you timestamped evidence so a person can make the call.

Early access

Make your next hire a real one.

We're onboarding hiring teams in small groups. InterviewShield is the first of Toolncy's interview tools, and waitlist members see the next ones first.

For hiring teams. No spam, unsubscribe anytime.