Is this photo AI?
Pictures made by AI are everywhere. Guess how good people are at spotting them, hunt for the classic mistakes in six pictures, then check one of your own photos for clues.
Spot the planted mistake
We drew six pictures in your browser. Some copy a mistake that AI image tools often make. Others are fine. Decide for each one, then see what we hid.
Check your own photo
Pick a picture. This page reads the hidden data inside the file, right here on your device: camera details, editing software, AI tags and content credentials. Then look at the picture yourself with our checklist.
On-device only: the file is read with your browser's own tools. Nothing is sent anywhere, and the preview is gone when you leave the page.
Your results will appear here.
Now look with your own eyes
Zoom in and tick anything you notice. Newer AI tools make fewer of these mistakes, so a clean picture is not proof that it is real.
Before you share it
Who posted it first? A real event usually has several photos from different people and angles. One perfect picture with no source is a warning sign.
A reverse image search can show older copies, the original before editing, or a fact check that already exists.
Some cameras and tools add a signed history to the file, called content credentials. When it is there, a credentials checker shows who made the picture and how it was changed.
Messaging apps and social networks often strip hidden data, and anyone can add fake camera details. Missing data proves nothing, and present data can be forged.
Sources: Nightingale, S. J. and Farid, H. (2022), "AI-synthesized faces are indistinguishable from real faces and more trustworthy", Proceedings of the National Academy of Sciences 119(8), and Lancaster University's summary (15 February 2022): 315 people sorted 128 faces into real or AI-made with 48% accuracy (chance is 50%); 219 people who were trained first reached 59%; 223 people rated the AI faces 7.7% more trustworthy. IPTC NewsCodes, "Digital Source Type" vocabulary (cv.iptc.org/newscodes/digitalsourcetype): "trainedAlgorithmicMedia" means "created using Generative AI", "compositeWithTrainedAlgorithmicMedia" means "edited using Generative AI", "digitalCapture" means captured from real life with a digital camera. Coalition for Content Provenance and Authenticity (C2PA) Specification 2.1: content credentials are stored in a JUMBF box, inside APP11 segments in JPEG files and a "caBX" chunk in PNG files. The planted mistakes (hands, writing, matching pairs, shadows, warped lines) are common examples given in media-literacy guides; they are clues, not proof. This page only reads the file; it cannot check a credential's signature, and it cannot tell for sure whether a picture is AI-made. Checked October 2026.