How to Spot an AI-Generated Photo or Video

AI-generated images used to be easy to spot. Blurry faces, extra fingers, nonsense text in the background. The newest ones are much better, and the fakes showing up in your feed, your group chats, and even news stories are good enough to fool busy people. You do not need special software to protect yourself. You need a few habits and a short list of tells.

Start with the context, not the pixels

Before you zoom in, ask basic questions. Who posted this? Is it an account you recognize, or a brand new page with no history? Does the caption make an extraordinary claim? Do other reliable sources show the same thing? Most fakes fall apart at this step because they travel without a credible source. If only one random account is posting a shocking image, that is the biggest red flag of all.

Tells in AI-generated photos

Hands are still the classic giveaway. Count the fingers, look at how they bend, and check whether a hand seems to melt into an object it is holding. Then look at text. Signs, labels, and logos in AI images often contain garbled or misspelled words, or letters that morph into decoration. Skin is another one: AI faces can look oddly smooth and waxy, like a plastic mask, especially around the cheeks and forehead. Check the background too. Earrings that do not match, glasses with uneven arms, repeating patterns that break, shadows falling in impossible directions. None of these proves anything alone, but two or three together should make you skeptical.

Tells in AI-generated video

Video is harder, but the fakes still glitch. Watch the mouth during speech: lip sync that drifts, or teeth that shimmer and change shape. Look at hands again, and at edges where a person meets the background, which can wobble or blur. Movement can look too smooth, like everything is floating slightly. And listen: AI-cloned voices often have a flat, even rhythm with odd pauses, and background noise that does not quite match the scene.

Use reverse image search

When a photo feels off, run it through a reverse image search like Google Lens. If the same image appears years earlier in a different context, or only on sketchy sites, you have your answer. For video, try searching a distinctive phrase from the caption plus words like "fake" or "debunked." Someone has often already checked.

What to do when you are not sure

Do not share it. That is the whole rule. Sharing a fake, even to debunk it, helps it travel. If it matters, check a source you trust, look for the original poster's history, and wait. Real events get confirmed by multiple outlets. Fakes stay lonely.

Why this matters more every month

The tools get better on a steady schedule, which means "I will just be able to tell" is not a plan. The plan is habits: check the source, look for tells, reverse search, and do not share what you cannot verify. None of that takes more than a minute, and it is the difference between being informed and being used.

Previous
Previous

Passkeys Explained: Time to Ditch Your Passwords

Next
Next

What AI Agents Actually Do (and How to Try One Safely)