People about to post
A quick look at an export before it goes to a platform, to see what it will carry with it.
Metadata viewer · AI metadata checker · No upload
View supported EXIF, XMP, IPTC and Content Credentials in your photo. Inspect camera and editing data, then choose whether to remove metadata from the downloaded copy.
No file loaded
Inspector
Waiting for file
Step by step
Drop a photo into the checker. It is parsed in your browser, so even private images stay on your device.
Red rows are flagged by the scanner. Open a populated row to read its available summary. You can strip them from here if you want to.
If you cleaned the file, save the copy and post that one. No flagged blocks does not guarantee privacy or predict a platform label.
Read the file before you remove anything
Use this photo metadata viewer to inspect supported blocks before deciding what to keep. The EXIF viewer summary shows selected device, software and GPS-presence information, not every EXIF field or the coordinates themselves. Expand a populated row for the summary this tool can extract.
The byte count is the size of the detected block, not an AI probability. Red means the scanner flagged a packet using its rules. “Not present” means no supported block of that type was found; it is not proof that a photo is authentic or contains no other identifying information.
A camera or phone can write device, capture and location details; editors can update them. The summary may show Make, Model, Software or GPS presence. To remove EXIF data for privacy, choose Everything after reviewing the other blocks that will also be removed.
Editors can store CreatorTool, edit history and descriptive fields here. A displayed DigitalSourceType describes the recorded source, not an independent judgment about the pixels. Removing a packet can also remove unrelated fields inside it.
Editorial tools can add captions, credit, keywords and source information. This row identifies a supported IPTC / Photoshop block; it is not a complete field-by-field reader. Keep attribution you still need before removing it.
Content Credentials can record provenance for captured or edited media. This scanner flags supported manifest blocks but does not verify signatures or display their full history. Their presence alone does not prove an image was AI-generated.
PNG text chunks and JPEG comments can contain notes, prompts or workflow data. The viewer shows chunk names and selected text summaries; it does not decode every compressed text payload.
Scanning does not overwrite the original. To remove metadata from a photo, choose Everything; to retain unflagged blocks, choose AI tags only. Download a separate copy and inspect it again.
Scope
False positives
That is the most common result people get here, and it does not mean the checker is wrong or that your photo is fake. One of these edits wrote the tag.
Check another file ↑Edit you made
What gets written to the file
Photoshop
compositeWithTrainedAlgorithmicMediaShows in the XMP row as a DigitalSourceType value.
Photoshop, Luminar
compositeWithTrainedAlgorithmicMediaLook for the same source type, sometimes alongside a CreatorTool entry.
Lightroom
C2PA manifest · c2pa.editedAppears as a Content Credentials row; the scanner does not read its contents.
Lightroom, Photoshop
C2PA manifest · trainedAlgorithmicMediaMay appear in both the Content Credentials and XMP rows of one file.
Who uses it
A quick look at an export before it goes to a platform, to see what it will carry with it.
Check what a delivered file records about how it was made and who is credited.
A first look at the metadata of a submitted image. It is one input, not proof of how the image was produced.
See what your phone and your editing apps write into every photo you take.
Tools and guides
FAQ
Using this on work that is not yours, or on AI-generated images? Read the disclaimer first.
It opens an image file and lists the metadata stored in it, highlighting fields that indicate AI generation or AI-assisted editing. It shows supported file signals, not a platform’s internal decision.
Not reliably, and no metadata tool can. It can tell you the file claims AI involvement. An AI image with its metadata removed will show as clean, and a real photo edited with an AI feature will show as flagged.
Either you checked a different copy from the one you uploaded, or the platform is using its own detection or a pixel watermark. Check the exact file you posted. If that file is clean, metadata is not the cause.
Yes. The viewer opens HEIC and AVIF files and shows the same rows as for a JPG: the camera make, model and software from EXIF, whether a GPS position is stored, and any XMP or Content Credentials data. A browser that cannot display HEIC will show a placeholder in place of the picture.
Yes. The file is read by code running in your browser and is never transmitted. The checker works with your connection turned off.
Yes. After a scan, choose whether to remove only the AI tags or everything, then download the cleaned copy from the same panel.
An IPTC field that states how an image came to exist. Values include “digitalCapture” for a camera photo, “trainedAlgorithmicMedia” for a fully generated image and “compositeWithTrainedAlgorithmicMedia” for a photo with generated elements. Platforms read it to decide on labels.