Instagram is mislabeling ordinary photos as AI-generated while letting actual synthetic images slip

through unmarked The "AI Content" badge that appears on Instagram is meant to signal at a glance that an image was created or meaningfully altered with generative tools. Over the past few weeks users have reported the system spinning out of control: camera photos that received nothing more than a background removal in Canva or a trivial blemish fix are being tagged, while images produced by generative models pass without a label. The result is a feed where nothing feels trustworthy.

This feels like déjà vu from February 2024. Back then the "Made by AI" label was attached to photos whose Adobe metadata merely indicated minor generative retouching — a single stain removed — even though the result was still fundamentally a photograph. Meta promised at the time to calibrate its approach so it would reflect the "amount of AI in the image," explaining that it relies on "industry-standard indicators" that partners embed in their tool outputs. Since then the company has published no updated detail on what those indicators are, how often scanning occurs, or how detection works. Meta did not respond to a request for clarification.

The core confusion stems from a distinction Meta does not make in practice. Background removal and object selection run on assistive machine learning (assistive ML) — technology Photoshop has used for more than a decade — not on text-to-image models that conjure new pixels from nothing. When the system treats both as "AI content," it creates noise that obscures the real problem: generative fakes that mimic reality.

Canva acknowledges the glitch, but users still see labels

Content strategist Jess Bruno reported that Canva confirmed some of its assistive AI tools had been "tagged as generative" by mistake and said the issue was fixed. The help page for Canva's background remover now states the tool "does not add Canva-developed AI content metadata to your design," though it is unclear when that note appeared. Canva did not respond to a request for comment. Despite the statement, users on Threads report the label still appears after using the background remover, while others note that images edited before the fix were never tagged at all — and also lacked the C2PA metadata Meta says it relies on.

Independent testing: only Meta AI triggered the label

The author ran a series of independent tests and found the badge was applied only to images edited or created entirely within the Meta AI app. There are also reports of mislabeling in cases where Canva's background remover was not involved at all, suggesting additional, unidentified factors are triggering the mechanism.