Instagram AI Labels Misfire in Both Directions
Meta's AI transparency push is undermined by a detection system that flags ordinary editing tools while missing embedded AI metadata, leaving the label over-applied to human work and under-applied to actual AI content.
Reporting from 1 source: GIGAZINE.
Instagram's AI content labels are misfiring in both directions, according to reports since July 2026. Creators who use editing tools such as background removal or layer creation see human-made work flagged as AI, including a 100 percent hand-drawn piece with no appeal path, while The Verge found images carrying C2PA and SynthID metadata went untagged. The only signal that reliably triggered the label was Meta's own generative AI output.
Meta renamed the label from "Made with AI" to "AI info" in July 2024 after photographers complained that ordinary Photoshop edits triggered it. The current wave of misfires, reported since July 2026, runs the other way too: Corkscrew Curiosities, a Kentucky antiques shop, says a single background removal pass or a "create layer" action flags work that took hours, and one user reports a 100 percent hand-drawn piece was labeled with no way to appeal.
The Verge ran its own tests and found images carrying C2PA and SynthID metadata went untagged, while a fresh account left to behave like a bot farm drew no enforcement for over two weeks. The only signal that reliably triggered the label, the outlet concluded, was content produced by Meta's own generative AI systems.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.
Sources
- GIGAZINE InstagramのAI検出ラベルが誤作動しまくっていると話題に