Instagram is mislabeing some real photos as AI content

In short: Instagram’s “AI Content” labels are being applied inconsistently, with some normal photos getting flagged and some AI-made images not getting flagged.

What's going on

Instagram uses visible labels to tell people when a post was made with generative AI, meaning tools that can create or add new parts of an image. Lately, users have reported that Instagram is putting an “AI Content” label on photos they say were taken normally and only lightly edited.

Several reports point to common editing tools as possible triggers. People say the label can appear after using Canva’s Background Remover, or after tiny touch-ups like removing a small speck. These tools can use AI in a basic “helper” way (like scissors that automatically cut out a subject), but they are not the same as tools that invent new people or scenes.

At the same time, The Verge reports that some clearly AI-generated images are slipping through without labels. In its own tests, The Verge uploaded images made or edited with a range of tools, including Google Gemini and some Apple photo features, and Instagram did not label many of them. The only reliable trigger in those tests was content made with Meta’s own AI app.

Canva told a user that some of its assistive tools were mistakenly being marked as generative, and said it fixed the tagging. However, some users say the problem continues. Meta did not respond to a request for clarification, according to The Verge.

What to watch

If labels stay unreliable, people may stop trusting them, which defeats their purpose. Watch for Meta to explain what signals it checks for, or to change how Instagram decides when to add an AI label.

Source: The Verge AI