Instagram once again incorrectly labels conventionally edited photos as AI content
Instagram incorrectly labels conventionally edited photos (Canva Background Remover, skin retouching) as “AI Content”, while actual AI content often escapes detection. A similar problem is recurring after the incident in February 2024.
In recent weeks, Instagram users have reported that the automatic “AI Content” labeling system incorrectly labels photographs that were neither created nor edited using generative AI. According to The Verge, the label typically appears after using Canva Background Remover or making minimal use of skin retouching tools — in one case, removing a single blemish from a photo was enough for the entire image to be labeled as AI. Conversely, content actually generated by AI often escaped detection in tests by The Verge — of several images tested, only two were labeled, both created directly in the Meta AI app.
In its statement, Canva confirmed that some of its assistive AI tools (a common type of machine learning used for years in Photoshop, for example, rather than generative text-to-image AI) had been incorrectly classified as generative, and said it had fixed the problem on its end. However, some users on Threads report that the labeling persists even after the fix, while others say that their older photos edited with the same tool before the fix were not labeled. There have also been cases involving no use of a Canva tool at all — including a photograph “poisoned” with a tool designed to prevent AI model training, while the unpoisoned version of the same photo did not receive the label. The erroneous label also appeared on photographs from the cosmetics brand About Face (founded by singer Halsey), whose representative said the image was an iPhone photo only slightly edited in a photo app, without using AI.
The detection system works by scanning IPTC and C2PA metadata, but according to the article, Meta has long failed to communicate clearly exactly which signals it looks for. A similar problem with the “Made by AI” label appeared as early as February 2024, when Meta promised to adjust its approach to better reflect “the amount of AI used in the image”. Meta did not respond to a request for a statement from The Verge, and Canva also did not respond to a request for comment.
Why it matters
Incorrect labeling undermines the credibility of the “AI Content” label as a whole — if the system labels even conventionally retouched photos while also letting actually generated content through, neither users nor businesses can rely on the label as an indicator of the origin of the content. For content creators and brands on Instagram, this means a risk that their real photographs will be publicly labeled as AI-generated (and without a clear explanation), which may harm their perceived authenticity without an easy way to remedy the situation, because Meta does not disclose its detection methodology.
Two audiences, two different impacts
What this means
For individuals
Instagram users may find that a conventionally edited photograph (background removal, skin retouching) is automatically labeled as “AI Content”, while content actually generated by AI often escapes this label, so the label cannot currently be relied on.
For a business
Businesses presenting themselves on Instagram (e.g. the cosmetics brand About Face) may face an erroneous “AI Content” label that makes photographs created without generative AI look untrustworthy, undermining the credibility of the content without a clear way for Meta to remedy the issue.
Risks and complianceCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.