How to Tell If an Image Is AI-Generated (or AI-Edited)

AI images now look real enough to fool almost anyone. There's no single trick that works every time, but a few quick checks, used together, will catch most fakes. This guide walks you through them, with examples, and it's honest about the limits, including our own tool's.

1. Look closely: signs of an AI-generated image

Zoom in. Many mistakes only show up at full size.

• Hands and fingers: extra, missing, or merged fingers, or nails that blend into skin. Newer AI tools have gotten much better at hands, so a normal-looking hand proves nothing.
• Text and logos: garbled letters, words that almost make sense, or signs that seem to melt. An unfamiliar brand or product name is not a sign of AI; new products come out all the time.
• Skin and hair: waxy, poreless skin, or hair that dissolves into the background.
• Accessories: mismatched earrings, glasses that merge into the face, or jewelry that fades away.
• Backgrounds: warped lines, doors and windows that don't line up, people or objects that blur into each other, or strange repeating patterns.
• Lighting and shadows: shadows pointing different directions, or reflections in eyes, mirrors, or windows that don't match the scene.
• Too perfect: flawless, symmetrical faces with a glossy "stock photo" look.

2. How to spot an AI-edited photo

AI editing changes part of a real photo: smoothed skin, whitened teeth, reshaped faces or bodies, removed or added objects, or a swapped background. Signs to look for:

• One area looks different from the rest: sharper, smoother, or blurrier.
• Skin looks airbrushed while the background still has normal camera grain.
• Edges around a person or object look too clean, or have a faint halo.
• Straight lines in the background bend near a body, a common sign of reshaping.
• The lighting on a face doesn't match the rest of the scene.

Honest note: small AI edits are the hardest thing to detect, for people and for software.

3. Check where the image came from

• Do a reverse image search with Google Lens, Bing Visual Search, or TinEye. If the same photo appeared earlier somewhere else with a different name or story, that's a red flag.
• Ask for the original file, or for more photos from the same moment. Real events usually have several photos from different angles.
• Check the account. Brand-new profiles with few connections and only a handful of perfect photos are common in scams.

4. Look for labels and file information

• Content Credentials: some AI tools and cameras attach a digital "nutrition label" (the C2PA standard) that can show how an image was made. Some platforms display an "AI info" or similar label.
• Invisible watermarks: some AI companies hide watermarks inside the images they create, such as Google's SynthID.
• Camera details (metadata): real photos often include the phone or camera model and its settings.

The limits: screenshots, social media, and messaging apps often strip labels and metadata. Missing information doesn't mean an image is fake, and editing software can also fake metadata.

5. Use an AI image detector, the right way

AI image detectors look for patterns people can't easily see, like lighting consistency, textures, compression, and camera sensor noise. JayKorn IQ, for example, checks several of these signals and tells you whether a photo looks Real, AI Edited, or AI Generated, with the reasons behind the result. When the evidence is mixed, it says "Needs Further Review" instead of guessing.

For the most accurate result:

• Upload the original file, not a screenshot.
• When possible, avoid images that were filtered, re-saved, or downloaded from social media.
• Read the reasons, not just the label.

The limits: why no detector is 100% accurate

Be wary of any tool that promises perfect accuracy, including us. Results are estimates.

Real photos can be flagged as AI when:

• They use heavy filters, beauty mode, or portrait effects.
• The phone processed them heavily (for example, night mode or HDR).
• They're photos of a screen, poster, billboard, or ad. The ad itself may be digitally rendered even though your photo is real. We've seen this ourselves: a real photo of a phone store kept getting flagged because of a large product poster on the wall.
• They're very small or heavily compressed.

AI images can be missed when:

• They come from brand-new AI tools.
• Only a small part of a real photo was changed.
• They've been screenshotted or compressed many times.

Use a detector's result as one strong clue alongside the other checks. Never use a single result on its own to accuse someone.

AI photo scams: what to do if something feels off

AI-generated photos show up in romance scams, fake marketplace listings, fake emergencies ("I'm in trouble, please send money"), and fake profiles.

• Slow down. Scammers rely on urgency.
• Ask for a live video call. Someone who never will is a red flag.
• Verify through another channel: call the person or company at a number you already know.
• Never send money, gift cards, or crypto to someone you've only met online.
• Report it. In the U.S., report fraud to the FTC at ReportFraud.ftc.gov, and report the account to the platform.

Share this guide with parents and grandparents. Older adults are frequent targets.

Frequently asked questions

Spot the Artificial. Know the Real.