If you’ve noticed an unexpected “AI Info” or “Made with AI” label appearing under an Instagram post that started life as an ordinary photo, you’re not imagining it, and you’re not alone. It’s happening to real businesses posting real content, and the reason comes down to one small technical detail most people have never heard of.

It’s not Canva being singled out

Instagram hasn’t built a rule that says “flag anything from Canva.” What it’s actually reading is something called C2PA Content Credentials, an industry metadata standard that records what software touched a file and how, embedded invisibly inside the file itself.

And this is how it is affected if you use one of Canva’s AI-assisted features, for example, its AI photo editor, background generator, or image generator. Canva writes that provenance information into the file when you export it. Instagram then reads that metadata the moment the file is uploaded. If the AI tag is present, the label can appear, even when the AI-assisted element was small, subtle, or barely noticeable in the finished image.

Plain design work is generally fine

This is the part worth holding onto. Cropping a photo, adding text overlays, adjusting colours, or laying an image into a branded template inside Canva doesn’t, on its own, trigger the same metadata flag. It’s specifically the AI-powered features that write the tag, not Canva as a platform. A photo edited manually and exported straight from Canva should carry no more AI signal than a photo edited anywhere else.

Why false positives happen

Because this detection runs on metadata rather than on what the image actually looks like, it can’t distinguish between a fully synthetic image and a real photograph that had one small AI-assisted touch-up somewhere in its editing history. That mismatch is common enough that a small industry of metadata-stripping tools has sprung up just to deal with it.

It’s also worth knowing that once a label has been applied to a published post, it generally can’t be removed from that post. The only real fix is to delete it and repost a version exported without the AI-tagged metadata.

What this means in practice

A few habits go a long way:

Check before you post. If any AI-assisted feature in Canva was used anywhere in a design’s editing history, even briefly, it’s worth exporting a clean version using manual tools only before it goes to Instagram.

Keep authenticity-signalling content genuinely manual. Founder photos, behind-the-scenes shots, and client testimonials are exactly the content where an unexpected AI label does the most damage to the message you’re trying to send.

Know your workflow. If your team uses Canva’s AI tools regularly, it’s worth building a quick check into the publishing process, rather than discovering the label after a post has already gone live.

None of this means avoiding Canva’s AI features altogether. They’re genuinely useful for plenty of tasks. It just means knowing which parts of your toolkit carry a digital signature, so you can decide deliberately where that signature does and doesn’t belong.

By Elly Georgiou
Elly manages the delivery of social media, digital content, and marketing campaigns across Popcom’s client portfolio. She brings rigour and creativity to campaign execution and is the person who ensures the strategy runs on time and on brief.