AI image generators have become good enough that a fake photo can pass a casual glance — and that means the casual glance is no longer a reliable test. Whether you are checking a news photo, a product listing, or a viral social media post, a short visual checklist will catch most fakes without any special software. This guide covers what to look for, in order of reliability.
Start With Hands, Text, and Edges
The classic tells have improved, but they have not disappeared. Hands remain the most common failure point: count the fingers, check whether they bend in plausible directions, and look for fingers that merge into each other or into objects. Text is the second check — AI models still garble lettering on signs, labels, and clothing, producing words that look right at a glance but dissolve into nonsense on closer inspection.
Then look at edges and backgrounds. AI images often have areas that are slightly too smooth — skin without pores, fabric without weave — sitting next to areas of odd, smeared detail. Background faces in a crowd may melt into each other, and objects at the edge of the frame may be half-formed. Reflections are another weak spot: mirrors, windows, and water that show something inconsistent with the scene in front of them.
Check the Light and the Logic
Next, test whether the image obeys physics. Shadows should fall in one consistent direction from a single light source; if one person's shadow points left while another's points right, something is off. Reflections, as above, should match. Perspective is equally revealing: check that tiles, windows, and fences recede consistently, and that objects at the same distance are roughly the same scale.
Then test the logic of the scene itself. Does the person's clothing make sense for the setting? Are there objects that serve no purpose or appear twice in slightly different forms? AI models assemble images from patterns, not from understanding, so they frequently include details that look plausible individually but make no sense together — a watch on both wrists, a door that opens onto a wall, a crowd where everyone faces the same direction for no reason.
Verify Outside the Image
Visual checks catch most fakes, but the strongest verification happens outside the frame. Reverse-image search the picture to see whether it appeared elsewhere with a different caption or an earlier date. Check the source: does the account posting it have a history of real photography, or did it appear yesterday with a single viral post? Look for the original context — a real event usually has multiple photos from multiple angles, while a fake usually stands alone.
Also remember that no single tell is proof. Real photos can have odd shadows, motion blur, and compression artifacts, especially after being re-shared and re-compressed many times. Treat your checklist as a scoring system: one odd detail is a curiosity, three or four together are a strong signal, and a mismatch with outside evidence is close to conclusive.
Why it matters
As generated images flood feeds, marketplaces, and even news, the ability to pause and verify is becoming a basic literacy skill — as fundamental as checking a headline before sharing it. The checklist above costs less than a minute and protects against the most common deceptions: fake evidence in arguments, scam listings with invented products, and manufactured "events" that never happened. In a media environment where seeing is no longer believing, that minute of skepticism is the difference between being informed and being manipulated.
A VCTRIA evergreen guide.
