AI image models strip veils, exposing cultural blind spots in safety filters

The Guardian reports that certain AI image-editing models can remove veils and religious symbols from photos, revealing gaps in safety barriers and cultural-context understanding. The claim is based on a media report, not a peer-reviewed study, and the specific models and methods are not detailed. This matters because it highlights a systemic flaw in AI content moderation that could enable cultural insensitivity or abuse.