Generation details
Prompt metadata, model names, seeds, samplers and ComfyUI workflows may be stored as text or structured records.
Inspect and remove C2PA, EXIF, XMP and embedded AI metadata from JPG, PNG, WebP and HEIC images locally before sharing to keep prompts, AI model settings, locations and creative workflows private.
Local browser processing
The AI metadata cleaner runs on your device; image files are never uploaded to our server.
Original or converted output
Keep original avoids recompression; conversions are re-encoded and checked again locally.
Inspect before cleaning
Review removable, preserved and unknown fields before applying the safe cleanup policy.
Verified before download
Every cleaned file is rescanned for metadata, format, dimensions and display integrity.
Understand the file
See the generation, provenance and photo data hidden behind visible pixels before the AI Metadata Remover cleans it.

AI image metadata can include prompts, model names, seeds, samplers and workflow data, plus provenance records describing how a picture was created or edited.
Ordinary EXIF, XMP or PNG text may also carry camera, GPS, time and author details. The AI metadata cleaner inspects these structures locally, removes recognized fields and reports uncertain data instead of deleting it blindly.
The AI Metadata Remover can inspect supported metadata exported by ChatGPT, Adobe Firefly, Stable Diffusion and ComfyUI. Support depends on the C2PA, XMP, EXIF, JUMBF or PNG text structures present in each file.
Prompt metadata, model names, seeds, samplers and ComfyUI workflows may be stored as text or structured records.
C2PA Content Credentials can describe origin and editing history through embedded manifests and JUMBF structures.
Camera details, GPS metadata, capture time and author information can travel with the visible image.
Recognized embedded fields are removed, image data is preserved and uncertain structures are reported.
Metadata scope
Remove AI metadata and private image details while keeping data required to display the image correctly.

The AI metadata remover covers embedded records revealing how an image was generated, edited or attributed.
Photo fields can reveal the device, location, capture time or creator.
Metadata explained
Before you remove AI metadata, identify which family contains a field and what that information describes.
| Metadata type | What it can contain | Why inspect it | Cleaner handling |
|---|---|---|---|
| C2PA / Content Credentials | Signed provenance claims describing how an asset was created, edited and attributed. | Can expose creation tools, edit history and recorded actions beyond visible pixels. | The C2PA remover removes embedded C2PA and JUMBF structures. The AI Metadata Remover cannot erase external records or pixel-level signals. |
| EXIF metadata | Camera settings, capture time, GPS coordinates, orientation, device and software details. | Can reveal when, where and on which device a photo was created or edited. | The EXIF remover cleans private fields while preserving display orientation. |
| XMP metadata | Editing history, creator details, AI labels and application-specific workflow properties. | Can carry workflow information or duplicate fields stored in other metadata families. | The AI metadata cleaner removes supported XMP packets and rescans the result. |
| IPTC metadata | Captions, keywords, creator, credit, copyright and editorial fields. | Useful publishing data can also expose private or outdated information. | The image metadata remover removes supported IPTC captions, keywords, creator, credit and copyright fields. |
Use cases
Remove AI metadata, EXIF and private photo details before sharing, client delivery or publishing, all locally in your browser.

Before sharing AI-generated artwork with a client or collaborator, inspect and remove embedded prompts, model settings, seeds, ComfyUI workflows and supported provenance records without sending the file to our server.

Photographers and everyday users can remove EXIF GPS coordinates, capture times, device details, editing software and author information before sending a photo privately or posting it online.

Clean image metadata before adding files to a website, portfolio, marketplace or social post. Review what will be removed or preserved, then download a verified result in the format you selected.

Inspect and clean up to 10 supported images in one local batch. Download verified files individually or package the completed results in a ZIP for delivery.
4 steps
Choose images, inspect the findings, remove supported metadata structurally, then verify and download the cleaned files.

Select up to 10 supported images. JPG, PNG, WebP and HEIC files stay inside your browser.
Review how many fields will be removed, preserved or kept with a warning.
Keep the original format without recompression, or convert the batch to a supported output format locally.
The cleaned file is parsed and decoded again before individual or ZIP download.
Supported formats
Inspect and remove supported image metadata locally. Keep the original format without recompression or create a verified conversion in the format you choose.
For photos and standard web images.
Keep original preserves
JPEG scan data, ICC profile and display orientation.
For transparent assets and interface graphics.
Keep original preserves
Transparency, APNG animation and safe chunks.
For modern web images and animation.
Keep original preserves
Image payloads, ICC, alpha, animation and unknown chunks.
Honest limits
Metadata cleanup changes supported file structures, but it cannot erase pixel-level signals or records controlled by external platforms.
SynthID and other invisible signals embedded in image pixels are outside metadata cleanup.
Removing metadata cannot control how visual AI detectors or classifiers judge an image.
External Content Credentials, soft bindings and platform records remain outside the downloaded file.
Platforms may still apply AI labels or require disclosure under their own rules and applicable law.
Questions and answers
Review the local workflow, supported files and honest capability boundaries.
Inspect what is present, clean supported metadata and verify the result before downloading your image.