Why Instagram Shows "AI info" on Your Photo

Quick answer

Instagram shows "AI info" when it reads a C2PA content-credential record or an IPTC DigitalSourceType field embedded in your file by the app you used to create or edit it — not because a human reviewed your post. The label was called "Made with AI" until May 2024, when Meta renamed it to "AI info" to cover partial AI edits too, not just fully AI-generated images.

What triggers the label

Meta announced in February 2024 that it would read "industry standard AI image indicators" on upload across Facebook, Instagram, and Threads. In practice that means two machine-readable signals:

C2PA Content Credentials — a manifest written by the generating or editing tool, which can include a digitalSourceType value like trainedAlgorithmicMedia.
IPTC metadata — a similar field that non-C2PA tools sometimes write instead.

Meta joined the C2PA steering committee in September 2024, and continues to read these signals at upload. Tools known to write this metadata automatically include Midjourney, DALL·E/ChatGPT images, Adobe Firefly, Google's Imagen, and — increasingly — ordinary editing features like Photoshop's Generative Fill, which is why a photo that was mostly real but had one AI-assisted retouch can still get labeled.

"AI info" vs "Imagined with AI"

Meta uses two different labels, and they mean different things:

"Imagined with AI" — specifically for photorealistic images made with Meta's own AI image features.
"AI info" — the broader label (used from July 2024 onward) covering content made or edited with third-party tools, triggered by the metadata signals above.

Does the label hurt your reach?

As of this writing, Meta has not published evidence tying the "AI info" label itself to reduced distribution, and nothing in its public ranking documentation ties reach to the label specifically. A post can still underperform for ordinary reasons — the label isn't established as the cause on its own.

If the label appeared on a real photo

This does happen: a photo edited with an AI-assisted tool (even for something as mundane as removing a background object) can pick up the metadata and get flagged, even though most of the image is an unedited real photograph. Tapping the label shows Meta's explanation and, where available, an option to give feedback if you believe it's inaccurate.

How to remove the metadata before you post

If you'd rather Instagram not read that signal at all — for a genuinely AI-assisted image you don't want auto-flagged for unrelated reasons, or just to understand what's in a file before sharing it — AiEraser's View mode decodes the actual C2PA manifest (claim generator, software agent, signature) so you can see exactly what Instagram would read, and Remove mode strips it before you upload.

Frequently asked questions

Can I remove the "AI info" label from a post I already published?

Not directly by editing metadata after the fact — Instagram reads the file at upload time. You'd need to delete and re-upload a cleaned version, or use Instagram's own feedback option on the label if you believe it was applied incorrectly.

Does removing C2PA metadata mean Instagram definitely won't label my photo?

It removes that specific signal, but Meta also runs its own visual classifiers trained on AI-generated images, so a sufficiently AI-typical image could still get flagged independent of metadata. Metadata removal addresses the most common trigger, not every possible one.

Is labeling AI content on Instagram required by law?

In some jurisdictions, yes — the EU AI Act includes disclosure obligations for certain AI-generated content, separate from and stricter than Meta's own voluntary labeling system. Meta's automatic "AI info" label is not a substitute for a legal disclosure duty if one applies to you.

What's the difference between this and the EXIF data my phone adds?

Different metadata entirely. EXIF records camera settings, GPS, and timestamps — it's not what triggers the AI label. See our EXIF guide and C2PA guide for the distinction.