How to Spot AI-Generated Images and Deepfake Videos

AI-generated images and deepfake videos have become much more convincing and much more common. These are photos and videos created or altered by artificial intelligence — showing people, places, and events that either don't exist or never happened. Learning to recognize them is an increasingly important skill.

A deepfake video can make a politician appear to say something they never said. An AI-generated image can show a celebrity endorsing a product they've never heard of. Fake "news photos" of disasters, protests, or crimes can spread across social media in hours and be completely fabricated.

There are telltale signs to look for, though AI quality is improving constantly and some fakes are now very difficult to detect without tools.

For photos, check the fine details: hands often have wrong numbers of fingers or unnaturally bent joints. Backgrounds are sometimes blurry or contain impossible architectural elements. Text within images (signs, labels) is often garbled or unreadable. Lighting and shadows sometimes don't match across different parts of the image. Hair can look waxy or perfectly smooth in unnatural ways. Eyes may lack catchlights (small reflections of light) or look glassy.

For videos, watch for unnatural blinking patterns (too fast or too slow), mouth movements that don't quite match the audio, skin texture that looks smooth or plastic-like, and inconsistent head movement. Audio quality may differ from the video.

Reverse image searching is one of the most reliable methods — if a photo has been circulating under different captions or on multiple unrelated sites, it may be out of context or fabricated.

The best habit is a brief pause before sharing anything shocking or outrageous: does this seem too perfectly timed? Does it confirm something you already believed without question? Those are reasons to verify before sharing.