AI-generated photos often look convincing at a glance but fail on specific details: fingers are the wrong number or shape, text inside the image is garbled, and backgrounds repeat in unnatural ways. Detection is getting harder as AI improves, so checking the source and context of an image matters just as much as the visual checks.
A photo that looks like a real person, a real customer review, or a real news event can be generated by AI in seconds — for free, by anyone. Most of the time these images are harmless. But they are increasingly used in fake profiles, fraudulent product listings, and misleading content online. Knowing where AI images still go wrong can help you catch them before you act on them.
Count the fingers carefully
This remains one of the most reliable tells. AI image generators frequently produce hands with the wrong number of fingers, extra joints, fingers that merge together, or fingertips that taper in odd ways. In any image where a hand is clearly visible, count all the fingers.
Zoom in if you can. Hands in the background or partially hidden behind an object are especially prone to errors — the AI often generates something that looks like a hand from a distance but falls apart on closer inspection.
Read any text visible in the image
When AI generates an image that contains writing — a shop sign, a book cover, a branded product label, a headline — the text is almost always garbled. It may look like real letters at a glance but will spell nothing when you read it closely.
Partially visible words often have wrong letter combinations, reversed characters, or repeating sequences that form no real words. Real photographs capture real text. AI invents something that resembles text without actually being readable.
Look at ears, teeth, and facial symmetry
Human faces are slightly asymmetrical in real life. AI faces often look too symmetrical or too smooth — a kind of perfection that can feel slightly off without being easy to name.
Ears are a common failure point. They may merge with hair, have odd cartilage shapes, or look like an approximation of an ear rather than a real one. Teeth, when visible, sometimes blur into a uniform stripe of white rather than showing distinct, individual teeth with natural spacing. These details are small but noticeable once you know to look for them.
Check the background for repetition or warping
AI often generates backgrounds by blending and tiling textures. Look at any patterned elements in the scene: brickwork, fabric, floor tiles, bookshelves, or crowds of people in the background. In AI images, these patterns sometimes repeat in ways that real scenes do not.
People in a crowd background may blend into each other or share duplicated faces. Windows may have inconsistent or impossible reflections. A background that looks more like wallpaper than an actual place is worth a closer look.
Notice how shadows fall
Lighting in a real photograph is consistent. Shadows fall in one direction from a single dominant light source, or blend naturally when multiple sources are present. AI images sometimes place shadows incorrectly — a shadow pointing the wrong direction, or an object that casts no shadow when it clearly should.
Look at the shadow under a person's chin, beside their nose, next to their glasses frames, or around any 3D object in the scene. When the shadow does not match the apparent light source in the rest of the image, that is a meaningful clue.
Examine jewelry and glasses closely
Like in deepfake video, small accessories trip up AI generators reliably. A necklace chain may merge into the collar or appear solid where it should drape. Earrings are often asymmetrical — one clearly different from the other in shape or position. Eyeglass frames commonly distort where they cross over the eye, or the lenses show reflections that do not match any visible light source.
If the accessories in an image look slightly undefined, merged into the face or clothing, or just not quite right, that is worth noting alongside any other signs you have already spotted.
Ask where the image came from
Visual inspection is only half the job. Ask: is this profile attached to an account with real, organic activity — tagged photos, comments from other real-looking people, a consistent posting history? Does this product review include specific, believable details about the purchase, or just general praise?
Context checks catch what visual inspection misses. And as AI image quality continues to improve, context will become even more important than the visual signs.
A Honest Note: Detection Is Getting Harder
The checks above work well on images made with current AI tools, but they are not foolproof, and AI generators improve continuously. Images that would clearly fail the finger check a year ago now pass it. This does not mean the skill is useless — it means that building the habit of checking both visual details and source context is the longer-lasting protection.
No single check is a reliable detector on its own. But combining several of these signs with a look at where the image came from gives you a reasonable basis for healthy skepticism.
What to try next: These same instincts apply to video — see How to Spot a Deepfake Video for the moving-image version of this guide. If you want to test dedicated detection tools, AI Detectors Tested covers what actually works and where current tools fall short.
