How to Prove a Photo Is Real
Someone, or some tool, says your photo is AI. A score cannot be argued with, but it can be outweighed. This page is the evidence routine: the original file, its capture metadata, Content Credentials, your process, and an honest reading of what a detector score means.
Start with the file. Upload the original below for an independent detector reading, then use the metadata checker for the capture record. Not a screenshot, not a re-export: the first file off the camera or phone.
Drag & drop, paste, or click to upload
PNG · JPG · WEBP · HEIC — your image is analyzed privately and deleted
This is an estimate, not proof — the detector can be wrong, producing both false positives and false negatives. Don't rely on it as your only source; verify anything important. wedetect.ai is not liable for decisions made from these results. See our Terms.
1. Keep and share the original file
Everything else on this page depends on this. The original is the RAW file, the HEIC from your phone, or the first JPEG the camera wrote. A screenshot, a WhatsApp forward, an Instagram download or a re-export from an editor is a new file with the evidence stripped out.
If the argument is happening on a platform that strips metadata, offer the original by email or a file link. The willingness to hand over the original is itself persuasive; generated images have no original.
2. Show the capture metadata
A camera or phone writes the make and model, lens, shutter speed, aperture, ISO, date and time, and often the GPS location into the file at the moment of capture. Run the original through the EXIF viewer and share the table. Settings that agree with each other and with the picture (a fast shutter for the frozen motion, a wide aperture for the blurred background) are hard to fake convincingly.
Honest limit: EXIF can be edited, so on its own it is strong evidence rather than proof. Combined with the original file and the next two steps, it usually settles the question.
3. Content Credentials, if your camera or software signs
Newer Leica, Nikon and Sony bodies, and Lightroom and Photoshop with the feature turned on, attach a signed record of capture and edits. If yours does, the Content Credentials checker reads and verifies it, and that is the closest thing to proof that exists today. If your gear does not sign, that is normal; most cameras do not yet.
4. Show your process
- Neighbouring frames from the same shoot: a burst, the shot before and after, the outtakes. Generators do not produce a contact sheet.
- The edit file with its layers or the RAW with its edit history. The edited photo detector shows what the exported file itself records about editing.
- Behind the scenes: a phone snap of the setup, the location, the model release, the invoice. Timestamps that line up.
- Your body of work: the same style, subjects and gear over years is context a reviewer can weigh.
5. What a detector score actually means
An AI image detector reads pixel statistics and returns a probability, not a fact. Heavily stylised or retouched photos, HDR, film grain, illustration-like lighting and strong compression can all push a real photo toward the "AI" end, and every detector has a false-positive rate. Ours publishes its numbers and its known weak spots on the accuracy page, including the fact that artistic images are where it errs most.
So when a platform or a commenter waves a score at you: ask which tool, ask for its published false-positive rate, and answer with the original file, the metadata and your process. A single score against a chain of evidence loses.
Frequently asked questions
My photo was flagged as AI. What do I do?
Do not argue with a score. Gather the original file (RAW, HEIC, or the first JPEG off the camera), show its capture metadata, and show your process: other frames from the same shoot, the layered edit, timestamps. A score is a probability from pixels; a chain of evidence is proof.
Can EXIF data prove a photo is real?
It is strong evidence, not proof. EXIF can be edited, and it is stripped by screenshots and most platforms, so its absence means nothing. Camera make and model, lens, exposure settings, date, time and GPS that all agree with each other and with your story carry weight, especially alongside the original file.
Does a screenshot keep the metadata?
No. A screenshot is a new image made by your phone or computer; it carries none of the original's capture data. Share the original file, not a screenshot of it.
I edited the photo. Does that make it fake?
No. Cropping, colour, exposure and retouching are photography. Keep the unedited original and the edit file; Content Credentials from Lightroom and Photoshop can record the edit history, and the edited photo detector can show what the file itself says.
Why do detectors flag real photos?
Because they work from pixel statistics, and heavy editing, strong stylisation, HDR, film grain, illustration-like lighting and heavy compression can all resemble generated images. Every detector, including ours, has a false-positive rate, and any honest one publishes it.
Can wedetect certify that my photo is real?
No. No tool can. What this page and the checks on this site do is help you assemble evidence a person can weigh: the original file, its metadata, any Content Credentials, and an independent detector reading.
Related: my art was flagged as AI · metadata checker · EXIF viewer · how accurate the detector is