Recent reports from CCTV highlight that AI image tools can reconstruct QR codes that have been partially obscured by mosaic filters. This capability allows users to scan and read information from vouchers or images that were previously considered protected by simple digital masking.
Technical analysis shows that AI successfully restores QR codes if the masking only covers the outer perimeter or is applied too thinly to hide the underlying details. However, if the entire QR code area is covered with a thick, opaque layer that obscures all visual details, the AI cannot reconstruct the original data.
Experts explain that AI restoration relies on the fact that the core data area of the QR code remains partially visible. If the core data is completely covered, the restoration process fails. To prevent unauthorized scanning, experts recommend masking at least 90 percent of the QR code area.
QR codes utilize built-in error correction mechanisms and positioning points, which allow devices to read information even when the code is partially damaged or obscured. These features, while designed for reliability, can be exploited by AI tools to recover data from images that users assume are safely redacted.
