Bimta
  1. Bimta
  2. Guides
  3. How to remove AI info from an Instagram post

How to remove AI info from an Instagram post

"AI info" on Instagram comes from two different places, and only one of them lives in your file. This guide separates them, shows how to clean what you can clean before posting, and is honest about what no tool can promise.

Updated 2026-10-25 · Written by the Bimta team

First, know which "AI info" you mean

1 · Metadata inside your file. Generators (ChatGPT, DALL·E, Midjourney, Firefly) attach a signed C2PA manifest plus XMP or IPTC markers stating how the image was made. Instagram reads this on upload — it is the part you control, and the part this guide removes.

2 · The label Meta applies. Instagram shows an "AI info" label when it finds those provenance signals, when you disclose it yourself, or when Meta's own classifiers decide. Removing metadata removes the first trigger only.

Remove the metadata before you post

  1. Make a copy of your photo. Keep the original untouched so you never lose the source.
  2. Scan the copy. Drop it into the scanner and read what it carries: C2PA manifest, AI labels in XMP/IPTC, GPS, camera data. Nothing is uploaded — the scan runs in your browser, and you can download the scan report.
  3. Remove what you want gone. In the Instagram workflow pick "Remove everything" (the platform re-encodes uploads anyway) or "Keep capture settings" if you want camera EXIF to stay. Every cleaned file is re-verified through an independent parser before download.
  4. Post the cleaned copy. Upload that file, not the original. Once Instagram has ingested the original, cleaning your local copy changes nothing about theirs.

How to read a scan result

  • C2PA manifest found — the file carries Content Credentials; this is the strongest AI signal metadata can hold.
  • AI-generated / trainedAlgorithmicMedia — the manifest explicitly records generation by an AI tool.
  • Generator strings in EXIF or comments — some tools (Midjourney famously) write the prompt or their name into EXIF description fields.
  • No supported AI metadata detected — nothing was found in the metadata. That is not proof the image is not AI-generated.

The offline test

Any browser-based cleaner can be tested: load the page, then switch off your network, then scan and clean your photo. If it still works with the network tab empty, the file never left your device. Bimta works the same online and offline.

What no honest tool will promise

Metadata removal does not touch pixel-level watermarks such as Google SynthID, and it does not blind Meta's visual classifiers. Anyone promising "your post will never be labeled AI" is lying to you. Use cleaning as a privacy and disclosure choice, not as deception — where disclosure is expected, disclose it yourself.

HEIC note. iPhone photos are often HEIC. Bimta detects metadata in HEIC too, but for cleaning, iOS "Most Compatible" (JPEG) output or a PNG export gives the smoothest path.

Related: Do social media sites remove EXIF data? and What is C2PA?

Scan your photo first. See the exact C2PA manifest and AI labels Instagram would read — in your browser, nothing uploaded.

Check my photo

Common questions

Does removing metadata stop the AI info label?

It removes the metadata trigger. Meta can still label a post through its own classifiers or your self-disclosure, and no tool can promise otherwise.

Can I remove the label after posting?

No. The label lives with Meta's copy of the media. Clean the file before you upload it.

Does a screenshot remove AI info?

A screenshot drops the metadata (and the quality). It also drops nothing Meta's classifiers might see — same limits as cleaning, with worse image quality.