Free · Batch up to 30 · No upload

Remove C2PA Content Credentials from Your Photos Online

Runs in your browser · nothing is uploaded · pixels are untouched

  • C2PA Content Credentials
  • XMP & EXIF
  • PNG text chunks

Remove C2PA metadata and AI content credentials from any image in seconds.

Strip AI generation signatures (ChatGPT, Midjourney, Stable Diffusion, Adobe Firefly) before posting to Instagram, TikTok, or your portfolio. Free, private, no signup.

How to use this C2PA remover

Three steps, under 30 seconds, no account required. All processing happens in your browser - your image is never uploaded to a server.

Why you need to remove C2PA metadata

Instagram and Facebook now automatically label AI-generated images with "AI Info." The problem? Photos you took, edited, or made with AI assistance get mislabeled as fully AI-generated - because C2PA content credentials are embedded in the file.

C2PA content credentials contain generation tool, timestamps, and edit history. When Instagram reads them, it applies the AI label automatically. Even a small Generative Fill in Photoshop can add a C2PA signature that labels your entire post. The generic photo pass - EXIF, GPS and copyright without leading with C2PA - is Photo Metadata .

Three problems C2PA causes

What is C2PA metadata?

C2PA stands for Coalition for Content Provenance and Authenticity - an open technical standard backed by Adobe, Microsoft, OpenAI, Google, and others. It embeds "content credentials" in image files, recording whether and how AI was used to create or edit them.

Content credentials can include: the generating tool (e.g. "Adobe Firefly" or "OpenAI DALL·E"), creation and edit timestamps, a list of modifications, and cryptographic signatures that prove the provenance chain.

Most modern AI image generators embed C2PA by default: Midjourney, ChatGPT image generation, Adobe Firefly, Photoshop Generative Fill, and Stable Diffusion exports all carry C2PA content credentials unless you explicitly strip them.

What this tool removes - and what it doesn't

This session removes C2PA signatures , content credentials , and provenance data - and, by default, can finish XMP and EXIF in the same pass. The picture should look the same; the file is rewritten in the browser, so it is not a byte-for-byte copy.

Processing is 100% browser-based. Your image is never uploaded to our servers. Free, no watermark, no account required. Batch up to 30 images per session.

Same cleaner as the other two metadata pages: one pass can include C2PA, XMP, and EXIF. This page only changes which story we tell first.

C2PA vs EXIF vs XMP - what's the difference?

Not sure which metadata your photo has? Here's how C2PA, EXIF, and XMP differ - and which tool to use for each.

C2PA (Content Credentials) - Written by AI generators (Midjourney, ChatGPT, Firefly). Contains AI provenance, generation tool, content signature. Remove it to avoid "AI Info" labels on Instagram/Facebook. → This page - C2PA Remover

EXIF (Camera Data) - Written by cameras & smartphones at capture. Contains GPS coordinates, camera model, timestamp, lens & settings. Remove it to protect your location & device identity. → EXIF Cleaner

XMP (Adobe Metadata) - Written by Lightroom, Photoshop, Camera Raw. Contains edit history, adjustment params, keywords, copyright fields. Remove it to hide your editing workflow from clients. → XMP Remover

This page leads with C2PA because that is the signed provenance most social apps read first. The same browser session can still finish XMP and EXIF in one download. Open the EXIF Cleaner or XMP Remover only when you want that story first.

Inspect C2PA before removing

Not sure if your file has C2PA? Use the AI metadata checker to see content credentials read-only in your browser before you strip.

AI metadata checker - inspect C2PA →

C2PA before Instagram upload

Why a small AI edit can label the whole post, and how to strip the signed manifest from the still you will actually publish.

Remove C2PA before Instagram - why AI Info appears →

One browser session, three metadata layers

This page is focused on C2PA because that is the signed provenance Instagram and Facebook read first. The widget above is the same browser cleaner used on the EXIF and XMP pages: the default pass strips C2PA, XMP, and EXIF together. There is no EXIF off switch. Keep a private master if you need original camera fields. Optional Inject camera EXIF can write placeholder make/model/date after cleaning (JPEG only) - that is not your original EXIF.

You do not need three downloads for one file. Run the session once, then spot-check a sample in the metadata checker if you want a before/after. Batch up to 30 images so a carousel or client set stays consistent.

Which platforms read C2PA on upload

Not every app treats content credentials the same way. The common pattern in 2026 is: signed AI assertions become a disclosure label, even when the pixels are mostly a real photograph.

Which exports usually write C2PA

When you should keep C2PA

Newsrooms, brands under disclosure rules, and anyone delivering evidence-style files may need the manifest intact. C2PA exists so software can show who generated or edited a file. If a client, regulator, or platform requires that chain, do not strip it.

For personal social posts, shop listings, and portfolio stills where the label is a metadata false positive, removing the manifest before upload is the usual prep step. Use only files you own or have rights to.

If the AI label is still there after cleaning

Stripping C2PA only removes file-level provenance. It does not erase SynthID-style pixel watermarks or a platform’s visual classifier. If Instagram still shows AI Info on a cleaned file, the trigger is probably not the manifest. Re-export from the design app can also write a new C2PA block - clean the file you actually upload, not an older copy.

For carousel posts, every slide needs the same pass. One dirty frame can label the whole set.

Feed stills, Reels covers, and Stories

AI Info is not only a Feed problem. A Reels cover exported from CapCut or Photoshop can carry a signed manifest even when the video track is a camera clip. Stories that start as a Firefly or Generative Fill still get the same scan. Clean the still you actually attach, not only the long-form export.

Carousels are unforgiving: one leftover manifest on slide seven can label the whole set. Process the batch in one session so every frame matches. If you later swap a slide, clean that replacement too - Instagram rescans the file you upload, not the one you cleaned last week.

When you are unsure, inspect first

A phone JPEG that never touched Firefly or ChatGPT usually has no C2PA. A read-only pass in the metadata checker confirms that before you strip anything. If the checker shows a signed manifest, run this page on the file you will actually upload.

iPhone Photos and Android gallery apps sometimes re-export after a crop or filter. That second export can write a new credential even if you cleaned an earlier copy. Always clean the last file that leaves your device.

HEIC, screenshots, and forwarded copies

HEIC from iPhone is usually camera EXIF, not a C2PA manifest. If a platform rejects HEIC, convert first, then inspect the JPEG - some converters add software tags. Screenshots almost never carry content credentials; if Instagram still labels a screenshot, that is a pixel or account issue, not this container.

WhatsApp and iMessage often recompress. A forwarded copy may already have lost the manifest. The risky file is the original you download from ChatGPT or Firefly and upload yourself.

Limitations

This page is about C2PA. The same browser session can also strip XMP and EXIF - leave those options on if you want a clean social file in one download. It does not remove pixel watermarks such as SynthID or a platform visual classifier. If a label remains after a clean export, see the disclaimer . Use only files you own or have rights to.

FAQ

What is C2PA metadata?

C2PA (Coalition for Content Provenance and Authenticity) is an open technical standard backed by Adobe, Microsoft, OpenAI and others. It embeds "content credentials" in images - recording whether and how AI was used to create or edit them.

Why do my photos show "AI Info" on Instagram?

Instagram reads C2PA content credentials embedded in images. If the image carries an AI-generated signature (from ChatGPT, Midjourney, Firefly, etc.), Instagram automatically labels it as AI. Removing C2PA data before posting prevents that label.

Does removing C2PA reduce image quality?

Resolution is not changed on purpose, and the picture should look the same. The file is rewritten in the browser, so it is not pixel-identical.

Which tools write C2PA data?

Midjourney, DALL·E, Stable Diffusion, Adobe Firefly, Photoshop's generative fill, and most modern AI image generators embed C2PA content credentials by default.

Is removing C2PA legal?

Yes. Removing metadata from images you own is legal in most jurisdictions. Content credentials are not DRM - they're provenance information, and you have full rights to images you created or licensed.

Will Instagram still label my photo as AI after removal?

If the C2PA credentials are completely stripped and no other AI signatures remain (e.g. invisible watermarks), Instagram can no longer detect AI provenance from metadata alone and will not apply the AI label.

Do I need to remove C2PA from every photo I post?

Only if you want to avoid the AI label or hide your editing history. For personal photos taken on a camera or phone (no AI involved), there's usually no C2PA data to remove.

Does the tool work on mobile?

Yes. The entire removal happens in your browser, so it works on iPhone, Android, iPad, and desktop - no app download needed.

Is this a different tool from the EXIF and XMP pages?

No. Same browser cleaner, same default full pass. This URL explains C2PA first. Keep a master copy if you need original camera fields.

How to remove ChatGPT image watermark?

ChatGPT, DALL·E, and many OpenAI exports embed C2PA content credentials - what creators call an "AI watermark" is often this invisible metadata, not a visible corner mark. Strip C2PA and XMP with the free tool above (same browser engine as all our removers). For GPT-4o native downloads and social upload steps, see the ChatGPT Image page . Pixel watermarks such as SynthID are not removed - see metadata vs pixel watermarks .

Do I need to clean Reels covers separately?

Yes if the cover is a still export. The video track and the cover file are different uploads. Clean the JPEG or PNG you attach as the cover.

Related guides

Platform upload prep and C2PA background - useful after you clean a file or when labels persist.