That "Urgent" Video From Your Boss? Your Eyes Can't Catch the Fake — Here's What Can

Full Episode Transcript

The best deepfakes don't get caught on the face. A fake video can look absolutely perfect on a single frame — flawless skin, natural expression, eyes that move just right. It gives itself away in the tiny gap between one frame and the next.

Trusted by Investigators Worldwide

Run Forensic-Grade Comparisons in Seconds

Court-ready facial comparison reports. Results in seconds.

Get Started

7-day refund guarantee**

If you've ever gotten an urgent video message from

If you've ever gotten an urgent video message from your boss, or watched a clip of a politician saying something shocking, this is for you. Because the way you've been told to spot fakes — trust your gut, look for something that feels off — barely works anymore. And honestly, that's a little scary. But once you understand what detection software actually checks, you stop feeling powerless. So how does a computer catch a fake your eyes can't?

Let's start with light. In a real video, the lighting on someone's face stays consistent from one moment to the next. A shadow under the nose at one instant should sit in the exact same spot a fraction of a second later — because the lamp in the room didn't move. Deepfake software is brilliant at painting a believable face. But it doesn't always understand that light bounces off skin in predictable ways across time. So you get little glitches — a shadow that flickers, or a shiny reflection on the nose that vanishes when it shouldn't. Detection systems call this the invisible physics check. Your eyes glide right past it. The algorithm doesn't.

And this is exactly where most of us get fooled. The common assumption goes like this — if the face looks real and moves naturally, the video must be authentic. People believe that because the face is the hardest thing to fake, so it feels like the best proof. But that logic is backwards now. The face is precisely where modern deepfakes are strongest. Real detection ignores how convincing the face looks. It asks whether the laws of light and motion hold up across the whole clip.

The second clue lives in the eyes. A deepfake doesn't fail on the big stuff like the shape of the jaw. It fails when it tries to keep eyes blinking naturally across thirty or more frames in a row. Real eyes blink with a steady, physical rhythm. Fakes blink at odd intervals — or skip the smooth transition a real eyelid makes. For the rest of us, that means the giveaway isn't one weird moment. It's the pattern stretched across time.

The Bottom Line

Now the technical twist — and this one's uncomfortable. When a video gets compressed and shrunk down to load fast online, it loses fine detail. Those tiny forensic fingerprints that expose a fake? Compression scrubs them away. So a crisp four-K deepfake gets caught easily. But the same fake, squeezed through a social platform, becomes nearly invisible to detectors. Sit with that for a second. The platforms that compress video the hardest are also where fakes spread the fastest. The compression that makes videos load quickly is accidentally protecting the worst fakes.

So here's the real shift. A convincing video isn't the problem anymore. The question that matters is whether the entire sequence behaves like it came from one real camera, in one real room, at one real moment. And that's a question no human can answer just by watching.

Let me leave you with the simple version. Deepfakes don't get caught by checking if a face looks real. They get caught by checking if the light, the blinking, and the tiny details stay consistent from frame to frame. And the more a video is squeezed to load fast online, the harder a fake is to catch. Your gut feeling about whether a video is real was never built for this fight — but knowing what to look past makes you a lot harder to fool. The full story's in the description if you want the deep dive.

That "Urgent" Video From Your Boss? Your Eyes Can't Catch the Fake — Here's What Can