NYTimes shows how to spot a fake AI photo by looking for small errors

In short: The New York Times published an interactive that challenges readers to find signs that a realistic-looking photo was made by AI, not taken in real life.

What happened

The New York Times published an interactive on October 8, 2026, titled “This Photo Is an A.I. Fake. Can You Tell?” It shows a scene that looks like someone inspecting a parcel inside a FedEx Office. The Times says the image was generated using Google’s Gemini, which is an AI system that can create pictures from text prompts, not captured at a real FedEx location.

The interactive points out several small clues that can give AI images away. One is perspective, where lines that should narrow toward the distance do not line up consistently. Another is text, where small lettering looks smudged or has strange shadowing, like it was copied imperfectly.

It also highlights odd connections between objects. In the image, a lanyard seems to float because it does not actually attach to an ID badge’s plastic holder. There is also what looks like the base of a television without the rest of the TV, like a drawing where someone forgot to finish an object.

Why it matters

AI-made images can look convincing at first glance, and the mistakes can be scattered, not obvious. The Times also notes that none of these visual clues are final proof on their own. A safer habit is to treat image checking like checking a rumor, look closely, then also confirm where the image came from and what the surrounding context is.

Source: NYTimes