A video call is not verification

Why a video call feels like proof

Seeing feels like knowing. A face that smiles at your joke, a voice that answers your question, a room with a fan turning behind it: it all arrives at once and feels too rich to be staged. After weeks of text, a call feels like the moment the person becomes real.

Researchers have a name for the starting point. Timothy Levine's truth-default theory holds that people tend to believe others unless something specific makes them suspicious, and that this default usually serves us well. A friendly call gives suspicion nothing to catch on.

The face itself is a weak witness. A meta-analysis by Charles Bond and Bella DePaulo, pooling 24,483 judges, found people tell lies from truths 54% of the time, barely above a coin toss, and are more accurate judging what they hear than what they see. A call adds a lot of confidence and very little information.

What a call checks, and what it does not

To be fair to the habit, a call does rule some things out. Someone who has only stolen a stranger's photos, and will not show up live at all, is filtered out. That is worth something. It is just a much smaller thing than it feels like.

The camera points both ways

A video call is the only common check where the person being checked takes most of the risk. Your face, your room and your voice are on the other person's screen, and you have no idea what happens to them there.

That is exactly where blackmail starts. NCMEC, the US child-protection centre, describes offenders moving conversations to private video chat, trading on lines like I'll show you if you show me, and then threatening to share what they got. Its CyberTipline received more than 186,800 reports of online enticement, the category that includes sextortion, in 2023.

There is a quieter cost too. When on-camera proof becomes the normal way to show you are genuine, people learn to perform on demand: to show more, sooner, to someone they have never met, and to call it being safe. The people most eager for that rule are often the ones who want something from the camera.

Why this matters more now

The face itself can now be made. In December 2024 the FBI warned that criminals generate video for real-time video chats, posing as executives, police or other authorities. In early 2024 an employee at the engineering firm Arup in Hong Kong sent HK$200 million after a video conference where, police said, the colleagues on screen had been built from footage of earlier calls.

It happens at smaller scale too. In Kozhikode in July 2023, a retired man made a WhatsApp video call to confirm a former colleague was really asking for help. Police said the face and voice were faked, and he sent ₹40,000 before he rang the colleague's saved number.

Training our eyes will not close the gap. In a study by Sophie Nightingale and Hany Farid, people sorted real faces from AI-made ones with 48% accuracy, rising only to 59% after training, and rated the made-up faces as slightly more trustworthy than real ones.

What works instead: check first, call second

We think the fix is an order, not a ban. Identity should be checked once, separately, by something neither of you can stage in the moment: a verified profile, a number you already had, a person you both know. Then, if you both want a call, it can do its real job, which is chemistry.

1. Check who someone is before any call, through a source they do not control.
2. Offer the same check of yourself. Proof that runs both ways feels less like suspicion.
3. If you call, keep it short and keep it to things you would be fine with someone keeping.
4. If a familiar face asks for money on a call, hang up and ring the number you already have for them.

Someone who says no to a video call is not proving anything either way. Many people decline calls for their own safety. Ask for a different kind of check instead.

Questions