Remove C2PA Before Instagram - Stop AI Info Labels
August 20, 2026
What Instagram is actually scanning
Think of three layers:
- C2PA - signed claims about who generated or edited the file. This is the usual metadata trigger for AI Info on Instagram and Facebook.
- XMP - unsigned XML. Midjourney prompts, Lightroom recipes, and some Firefly notes live here. Instagram may not use XMP the same way it uses C2PA, but the packet still travels if you send the original elsewhere.
- EXIF - camera GPS, model, timestamps. Privacy, not the main AI-Info switch.
Removing only GPS will not delete a Firefly credential. Removing only C2PA will not delete a Midjourney prompt. Default on this site is one pass that can clear all three. Keep a master if a client still needs shooting data.
Newsrooms and brands that must disclose provenance should keep the manifest. This article is for social posts, shops, and portfolios where the label is a leftover assertion, not a legal requirement. Use only files you own or have rights to.
Covers, carousels, and the last export
The video track and the cover still are different uploads. A CapCut or Photoshop cover can carry C2PA while the clip itself looks like camera footage. Stories and carousel slides each get their own file. One dirty slide can label the set.
Phone Photos apps often re-export after a crop. Clean the last file that leaves the device, not last Tuesday’s download. Screenshots rarely contain a C2PA block; if a screenshot still gets AI Info, that is usually a pixel classifier or an account flag, not this container.
HEIC from a camera roll is usually EXIF, not a signed manifest. If Instagram rejects HEIC and you convert first, inspect that JPEG - some converters add software tags, and a later Firefly pass can add C2PA on top. WhatsApp and iMessage recompress; the risky file is the original you downloaded from ChatGPT or Firefly and uploaded yourself.
Strip the manifest, then post
When you are ready to upload, run a browser session on the still you will attach - same cleaner as the EXIF and XMP pages, this URL just leads with C2PA. Use the C2PA remover , download the full-strip file, and post that copy. Keep a master if you still need original camera fields.
If AI Info remains, the trigger is probably not the manifest. Invisible watermarks and visual models are out of scope. Re-exporting from Firefly after a clean writes a new credential; clean again.
Remove C2PA before Instagram upload
Strip the signed manifest from the still you will post, then upload that file.
- Export the still you will post - Save the JPEG or PNG from Firefly, ChatGPT, Photoshop, or CapCut - including Reels covers.
- Strip the C2PA manifest - Run a browser session on that file. The default pass is a full strip. Keep a master if you need camera fields.
- Upload the download - Create the Instagram post with the cleaned file, not an older export.
Frequently Asked Questions
Does Instagram read C2PA metadata?
Yes. On upload, Instagram can read a signed C2PA manifest. AI-related assertions often become an AI Info or Made with AI label on the post.
How do I remove C2PA before uploading to Instagram?
Export the still you will actually post, strip the C2PA manifest in the browser, then create the Instagram post with that download. The picture should look the same.
Will removing C2PA guarantee no AI label?
It only stops the metadata-driven label. Pixel watermarks, visual classifiers, or an older account flag can still show AI Info. See the site disclaimer.
Can I keep camera EXIF and still remove C2PA?
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.
Does Instagram also read XMP?
It can. Prompts and software tags in XMP sit in a different block from C2PA. A full pass clears both; C2PA is the signed trigger Instagram is known for. Remove AI Label Team