The Complete Guide to Spotting AI Generated Images and Videos

You’ve probably scrolled past a perfect sunset photo that felt just a little too perfect, or watched a video of a politician saying something that made you squint. AI-generated media is now so good that even experts sometimes hesitate. But before you start questioning every pixel, know this: AI makes mistakes that are brutally consistent. Once you know what to look for, you’ll spot them faster than you’d think.

The Anatomy of an AI Image

AI image generators work by predicting what should come next in a grid of pixels, like a supercharged autocomplete for images. This leaves telltale signs that are hard to erase.

1. Hands and Fingers Are Still Nightmare Fuel

AI hates hands. They almost always have extra fingers, missing thumbs, or digits that bend at impossible angles. Look closely at hands, especially if they’re holding objects or gesturing. A person with six fingers on their left hand and three on their right? That’s a dead giveaway.

Quick check: Count the fingers. If it’s not five, it’s fake. Even when AI gets the count right, the fingers often look like sausages or have strange, unnatural root-like connections.

2. The Background Is a Hallucination

AI is great at making a subject look convincing, but backgrounds are where it fails. Look for: - Disappearing text: A street sign that smears into gibberish, or a storefront sign that reads “Bakery” but the letters warp into a jumble halfway across. - Impossible geometry: Chairs that float slightly above the floor, windows that open into brick walls, or patterns on wallpaper that change mid-wall. - Missing or duplicate objects: A crowd where two people have identical faces, or a single tree that appears twice at different angles.

3. Lighting Doesn’t Add Up

Real photos have one main light source. AI often applies light from multiple angles simultaneously. Look for shadows that go in different directions, reflections on glasses that don’t match the scene, or a face that’s lit as if from above while the background is lit from the side.

Pro tip: Check the eyes. In real photos, the catchlights (reflections in the iris) should point toward the main light source. If you see two bright spots pointing different ways, something’s off.

Video: The Moving Lie

AI-generated videos are still in their “uncanny valley” phase, but they’re catching up fast. Here’s what to watch for.

1. The Skin Warp

When people move in AI video, their skin often looks like it’s made of elastic. Watch the edges of faces, especially around the jaw and cheeks. If the texture shifts or ripples like someone’s pulling at a wet cloth, it’s synthetic.

The blink test: Real people blink unevenly and half-closing their eyes. AI blinks tend to be too fast and perfectly symmetrical, like two car doors slamming at once.

2. The Resolution Drop Zone

AI videos often have a region around the face that’s crisper than the rest, because the model prioritizes the subject. Look for a sudden drop in sharpness as your eye moves from the main person to the background. If the background looks like a blurry watercolor painting, you’re looking at a generated frame.

3. Lip Sync Is Terrible

AI lip-syncing tools are getting better, but they still struggle with words that have “M,” “B,” and “P” sounds. Watch the mouth carefully during these consonants—they often look “mushy” or out of sync by a few frames. If the speaker’s jaw moves but the lips barely open, someone’s lying.

Audio: The Forgotten Clue

AI-generated voices have two tells: - Breathlessness: Real humans pause to breathe. AI voices often talk in a continuous stream with unnatural, robotic inhales placed at random intervals. - The “stutter-rhythm”: AI voices tend to have a subtly rhythmic cadence, like a metronome running under the words. Record a few sentences and play them back—if the pauses are too regular, it’s synthetic.

Practical Steps: What to Do Now

You don’t need a forensic lab. Here’s your quick-check workflow:

1. Zoom in to 200% on the image—look at edges, hair strands, and reflections.
2. Reverse image search with Google or TinEye. AI images often lack meaningful metadata, but real stock photos will have a history.
3. Check the EXIF data if possible. AI images may skip camera info entirely, but a blank “no data” field isn’t proof—smartphones can strip it, too.
4. Look for AI artifacts apps like “AI or Not” or “Illuminarty” can help, but they’re not perfect. Use them as a second opinion, not a verdict.
5. Scratch the background trick: In Photoshop or a free editor, clone a patch of the background. If the texture is too uniform or repeats like wallpaper, it’s likely AI.

The Bottom Line

AI-generated media is a mirror—it reflects what we expect to see, but in a funhouse way. The best defense isn’t a tool; it’s skepticism. If something feels too perfect, too surreal, or too convenient to be true, it probably isn’t real. Trust your gut, then check the hands. They never lie.