AI Shield

Detection is a habit, not a talent.

Nobody spots every fake. What works is knowing the few places these things still break, and what to do when something feels off.

One detail is never proof

A strange hand, an odd shadow, a word that reads slightly wrong. Any of these can appear in a real photograph, and any of them can be absent from a generated one.

Treat every sign below as a reason to check the source, never as a verdict on its own. The people who get caught are not the ones who missed a detail. They are the ones who never checked.

Where to look

Hands and teeth

Counting and repetition are still weak points. Look at fingers gripping objects and at teeth in a wide smile.

Text in the image

Signs, labels and screens often hold letters that almost spell something. Zoom in on any writing.

Edges and hair

Where a person meets the background, look for softness, smearing, or strands that end nowhere.

Light and shadow

Check that every shadow in the frame falls the same way, and that reflections match what is in front of them.

Faces in motion

Watch blinking, and watch the mouth against the sound. Small timing errors are easier to feel than to see.

Voices on a call

Cloned voices handle scripted speech well and unexpected questions badly. Interrupt and ask something only they would know.

The account behind it

Age of the account, comments turned off, no history. The source is usually louder than the content.

The ask

Almost every one of these ends in money, credentials or a code. That request is the most reliable signal of all.

The pressure

A deadline, a secret, a reason not to call anyone. Urgency exists to stop you doing exactly what this page teaches.

Try three

Three real patterns. Would you act, or check first?

Next: the five step system

Free checklist Full guide