How to Check Photo Metadata (EXIF) for AI and Edits

Learn how to check photo metadata and EXIF data to spot AI generation and edits, why it is easily stripped, and how to confirm with a free AI image detector.

To check a photo's metadata for AI or edits, open the file's EXIF and provenance data and look at the camera info, software field, and any C2PA Content Credentials. Real camera photos carry device details; AI or edited images often show editing software, generation parameters, or a signed AI manifest instead. The big catch is that metadata is easily stripped, so a blank result proves nothing, and you confirm with an AI image detector. Here is how to read image metadata and what it can and cannot tell you.

Free, no signup: Detect AI images · Check text for AI · Count words

Key Takeaways

• EXIF metadata can show the camera, date, settings, and editing software behind a photo.
• AI and edited images may reveal generation parameters, editing apps, or C2PA provenance data.
• Metadata is easily stripped by screenshots, social uploads, and re-saves, so absence proves nothing.
• Provenance standards like C2PA Content Credentials are signed and tamper-evident, unlike plain EXIF.
• When metadata is gone, confirm with a pixel-level AI image detector plus the visual tells.

What EXIF metadata is

EXIF (Exchangeable Image File Format) is data embedded in an image file that records how it was made. A photo from a real camera or phone typically stores the device make and model, the date and time, and shooting settings like aperture, shutter speed, and ISO. Editing software can add its own fields when a file is saved.

Think of it as the image's built-in record. When it is intact, it is a fast first clue about origin: a genuine camera capture looks different in the metadata from an image that only ever passed through editing or generation software.

How to read it

You do not need special skills to check EXIF. A few simple routes:

• On Windows, right-click the file, choose Properties, then the Details tab.
• On Mac, open the image in Preview, then Tools, then Show Inspector, and the info tab.
• Online EXIF viewers let you upload a file and see the full data.
• Provenance viewers read signed Content Credentials specifically.

Look at three things: the camera or device field (present on real photos, often blank on generated ones), the software field (may name an editor or AI tool), and any C2PA or "AI generated" declaration.

What metadata reveals about AI and edits

Metadata can carry surprisingly direct clues. Some AI tools and pipelines write generation parameters into the file, the prompt, model, sampler, or seed, which are a strong sign an image came from a generator rather than a camera. Editing apps often leave their name in the software field, showing a photo was processed.

Most importantly, provenance standards go further than plain EXIF. C2PA Content Credentials attach a signed, tamper-evident record that can explicitly declare an image was AI-generated and by which tool. OpenAI attaches these to its images, and other makers are adopting the C2PA standard. Unlike ordinary EXIF, a signed credential cannot be quietly forged, only stripped.

The big limitation: metadata is fragile

Here is why metadata alone is never enough. It is trivially easy to lose. A screenshot drops it entirely. Most social platforms strip it on upload to save space and protect privacy. A simple re-save can wipe or overwrite it. And plain EXIF fields can be edited by anyone, so a determined faker can change them.

This means two things. A missing camera field does not prove an image is AI, it usually just means the file was processed somewhere. And present metadata is only as trustworthy as its source, unless it is a signed C2PA credential. So treat metadata as a helpful clue, not a verdict, especially when it is absent.

When metadata is gone, read the pixels

Because most images you meet in the wild have stripped metadata, you need a method that does not depend on it. That is what a pixel-level detector provides. Upload the image to the free AI image detector, read the confidence score, and turn on the heatmap to see which regions carry generation artifacts.

This works regardless of metadata, because it reads the image content itself. So the practical workflow is layered: check metadata and provenance first, and when they are missing or inconclusive, fall back to detection and the visual tells. We explain the honest limits of each in do AI image detectors actually work.

A complete verification workflow

Put it together into a repeatable check:

1. Open the file's EXIF and look at the camera, software, and date fields.
2. Check for C2PA Content Credentials in a provenance viewer, which are signed and reliable.
3. If metadata names an AI tool or shows generation parameters, that is strong evidence.
4. If metadata is missing or ambiguous, run the image through the AI detector and read the heatmap.
5. Add the visual tells and, for high stakes, a reverse image search to see prior appearances.

No single step is proof on its own, but together they build a confident answer. We compare two of these steps in reverse image search vs AI image detector, and cover the provenance side in what are C2PA Content Credentials.

What real camera metadata looks like

Knowing the "normal" pattern helps you spot the abnormal. A genuine photo straight from a phone or camera usually carries a rich set of fields: the device make and model, the lens, the exact date and time, GPS coordinates (if enabled), and shooting settings like aperture, ISO, and shutter speed. That density is hard to fake convincingly, because a real capture writes all of it automatically.

An AI-generated image, by contrast, typically has no camera or lens data at all, because no camera was involved. Instead it may carry a software field naming a generator, embedded generation parameters, or a C2PA manifest declaring AI origin. So the absence of any camera data on an image presented as a real photo is a yellow flag worth following up, though remember it can also just mean the file was processed and stripped.

Metadata is a clue, not a conviction

The recurring theme is worth stating plainly: metadata is one input, never the whole verdict. Present camera data can be edited or spoofed. Missing data usually just means the file passed through a platform that stripped it. Only signed provenance like C2PA is tamper-evident, and even that is often gone. So a responsible check never rests on metadata alone.

That is exactly why the layered approach matters. Read the metadata for what it honestly offers, a fast clue when it survives, then confirm with pixel-level detection, the visual tells, and a reverse image search. The combination is what turns a hunch into a confident call, especially on the many images where the metadata has already been wiped by the time they reach you. Used that way, metadata becomes a useful first step in a bigger process rather than a false sense of certainty on its own. Check it when you can, trust the signed credentials most, and let the detector carry the weight whenever the file has been stripped bare.

Confirm any image free, no signup: Detect AI images · Check text for AI · Count words

FAQ

Can EXIF metadata tell me if an image is AI-generated?

Sometimes. Some AI tools write generation parameters into the file, and C2PA Content Credentials can explicitly declare AI origin. But metadata is easily stripped, so a blank result does not prove an image is real.

How do I view a photo's EXIF data?

On Windows, right-click and open Properties then Details. On Mac, use Preview's Inspector. Or upload the file to an online EXIF or Content Credentials viewer.

Why is the metadata missing from most images?

Screenshots, social media uploads, and re-saves strip EXIF and provenance data. That is normal, so absence of metadata is not evidence either way.

Is EXIF data reliable proof?

Plain EXIF can be edited, so treat it as a clue, not proof. Signed C2PA Content Credentials are tamper-evident and far more trustworthy when present.

What if there is no useful metadata?

Run the image through a free AI image detector, which reads the pixels rather than the file data, and add the visual tells and a reverse image search. GPTOne's detector is free with no signup.

GPTOne's AI image detector is free with no signup and reads the pixels when metadata is gone, at gptone.me.