AI Image Quiz — Real or AI?
AI image generators like DALL-E, Stable Diffusion, and Google's Imagen are getting incredibly good at producing fake vs real images that are nearly impossible to tell apart. Take the free AI detection test — then learn the tells so you never get fooled again.
How Does the AI Image Quiz Work?
Each round presents you with 4 images — but only one is real. The other three were created by modern AI image generators. Your task: find the authentic one.
Choose from multiple categories including famous paintings, iconic buildings, animals, and food. After each guess, we reveal which image was real and which ones were generated by AI.
Choose Your Category and Start the Quiz
Pick a category below to jump right in
More quiz modes
Why AI Image Detection Matters in 2026
In 2026, AI-generated images are no longer a novelty — they appear in news articles, product listings, social media profiles and political campaigns. Tools like DALL-E 3, Flux, and Google's Imagen can generate a photorealistic image in under 10 seconds at zero cost.
The ability to distinguish fake vs real images is becoming as fundamental as reading critically. This free AI detection test is designed to train that skill — one round at a time.
Can You Spot AI-Generated Images? A Breakdown by Tell
🖐️ Hands and Fingers
The single most reliable tell across all AI models. AI struggles to consistently generate five correctly-articulated fingers — you'll often see six fingers, fused knuckles, or fingers that bend the wrong way.
What to look for: Count the fingers. Check if knuckles are on the correct side. Look for hands that disappear into pockets or clothing suspiciously.
📝 Text in the Image
AI models generate text as visual patterns, not as readable characters. Shop signs, labels, t-shirt prints and book covers in AI images are typically garbled, misspelled, or styled like letters without meaning anything.
What to look for: Zoom into any signs, logos or labels. If the text looks like letters but doesn't spell anything, it's almost certainly AI.
🌿 Backgrounds and Edges
AI composites the subject and background separately and blends them together. This often creates a "soft halo" around subjects, especially around hair or fur. Backgrounds can also contain objects that are logically impossible or architecturally wrong.
What to look for: Look at the transition between the subject and background. Does hair blend unnaturally into the background? Are distant objects slightly melted or dreamlike?
🪞 Symmetry and Proportions
Real world objects — faces, buildings, animals — have subtle asymmetries that AI tends to eliminate or over-exaggerate. An AI face can look slightly "too perfect". An AI building might have windows that are mathematically uniform in a way no real architect would design.
What to look for: Are the ears the same size? Are all windows identical? Does the face feel like a mirror image of itself?
🧵 Textures
AI excels at generating plausible-looking textures at a glance, but they break down under scrutiny. Skin is often too smooth with no visible pores. Fabric folds repeat in unnatural patterns. Food textures — especially things like bread, meat or cheese — look correct from a distance but have no real depth.
What to look for: Focus on one texture area. Does it look the same everywhere with no variation? Real textures have imperfections.
💡 Lighting and Shadows
Every real scene has a single consistent light source (or a clear combination). AI images frequently have lighting that contradicts itself — a subject lit from the left while their shadow falls to the left as well, or reflections in eyes that don't match the scene's light.
What to look for: Where is the light source? Do the shadows on the subject match? Do eye reflections make sense?
Fake vs Real Images: How AI Fails by Category
Different quiz categories expose different AI weaknesses. Here's what to look for in each:
AI fur looks correct at a glance but unnatural up close — strands often merge into blobs. Eyes are too reflective or symmetric. Animal anatomy (joints, paws, tails) is frequently wrong.
AI food photography looks delicious but sterile. Real food has drips, crumbs, imperfect cuts and natural lighting variation. AI food is often too symmetrical and too clean.
Architectural details like windows, columns and ornaments repeat too uniformly. Perspective lines often don't converge correctly. Real landmarks have visible wear, shadows and irregular details.
AI imitations of famous art styles are impressively accurate — but brush stroke direction is often inconsistent, and signatures or text in the painting are garbled or missing entirely.
Which AI Models Are Used — and What Gives Them Away?
Not all AI image generators make the same mistakes. Here's a quick breakdown of the models you'll encounter in the quiz:
- DALL-E 3 — consistent quality, but slightly over-saturated colors and overly clean compositions
- Stable Diffusion / FLUX — highly realistic, but struggles with text, hands and complex scenes with multiple subjects
- Midjourney — strong artistic style, often too cinematic or "beautiful" to be a casual real photo
- Imagen (Google) — excellent photorealism, but backgrounds can have subtle warping
- Ideogram — best at text in images, but overall composition can feel staged