Has this photo been edited?

People often arrive at this page with a suspicion: a client sent a photo that looks slightly off, a marketplace listing seems too good, or a picture was passed around and something about it changed. The useful question is not “is this fake?” — no browser check can answer that — but “was this copy re-saved or stripped along the way?” That is a question a file can genuinely help with.

This photo tampering check reads the structure of a JPEG and the metadata it still carries. It reportssignals, not conclusions. Each signal comes with the everyday situation that produces it, because most of these traces show up on perfectly ordinary files.

The five signals this check looks at

Every item below is a heuristic — a hint worth reading, not a verdict. None of them measures how likely it is that the picture was altered, and this page deliberately reports no percentages.

1. Signs it was saved more than once. JPEG files can keep a small preview image, a comment block, or more than one metadata package. When two or more of these are present, the copy was probably written to disk by more than one pass of software. When this is normal: almost every export, download, and resize rewrites the file, so this is common and usually means nothing more than a re-save.
2. Camera and software records disagree. A file can name a camera and also name editing software, or claim an image size that does not match the picture actually in the file. That is worth reading carefully.When this is normal: photo apps, RAW converters, and batch compressors often keep the camera name while updating the software field or resizing the image.
3. Size and compression look software-made. A picture whose dimensions match a common export preset, and whose compression matches a standard software encoder, looks like a clean export rather than a straight-from-camera shot. When this is normal: many cameras make tidy sizes, and nearly every image published online is exported with standard settings.
4. Unusually different detail in one area. Comparing small regions of a picture can reveal a spot that was retouched or pasted in. This browser build does not run that comparison; it needs the server version of the check.When this is normal: night scenes, heavy grain, phone computational photography, and screenshots all make some areas differ without any manual edit.
5. No capture record on a re-encoded file. A JPEG with no camera or capture-time fields, standard compression, and a moderate size has the everyday shape of a file downloaded from a social platform or messaging app. When this is normal: many people switch off location and metadata in their camera or export settings, which produces exactly the same clean file.

How to read a result

The result groups what it found into “worth a second look” and “common in normal files.” A signal in the first group is simply less common, not more damning. A signal in the second group appears so often after a normal export that it usually tells you nothing about the picture itself. Every card states the finding, what it can mean, and the situation that explains it away.

Read the card before you quote it. “The stored image size does not match the picture” is a useful, specific statement. “This photo is doctored” is not something this check can say, and it would not be fair to read it that way.

Why an empty result is not a clean bill of health

A quiet result means this check did not see these particular traces in this copy. It does not mean the file was never edited, and it certainly does not mean the content is genuine. Screenshots, format conversions, messaging apps, and social platforms routinely rewrite images and remove the very records these signals are built from. If the history matters, ask for the original file the creator saved, and compare that copy instead.

The check also says nothing about who made the content. Editing traces and AI generation are different questions. A heavily retouched photograph and an AI-generated picture can look similar to a structural check, and this tool does not try to separate them. If you want to read the other kinds of evidence a file may carry, start with theEXIF metadata viewer for ordinary camera and software fields, or theContent Credentials checker for a signed provenance record.

What this check cannot do

It cannot recover a record that was stripped before you received the file. It cannot read proprietary invisible watermarks such as Google SynthID — that question belongs withGoogle’s official detector. And it cannot turn a handful of small traces into a confident story about what happened to the picture. For the honest limits of file checking, seewhat can be detected in AI media.

This site only reads files. It does not strip metadata, remove watermarks, or offer any way to cover a file’s tracks. If that is what you are looking for, the answer is a firm no, andthis page explains why. The goal here is to help you understand what a file says about its own history, not to help anyone disguise it.