GeoEdit protects photo location privacy by subtle image changes
GeoEdit: Adversarial De-Evidencing for Silent Geoprivacy Protection via Image Editing
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
Photos can reveal where they were taken even if location tags are removed, because smart AI models recognize small clues in the images. The authors developed GeoEdit, a method that makes tiny edits to parts of a photo that give away its location. These edits confuse AI models' guesses while keeping the picture looking natural to people. GeoEdit learns which parts to change and how, making it easier to protect your privacy silently.
What this means in practice
- •For mobile app developers: Integrate GeoEdit to automatically modify user photos for location privacy before sharing on social platforms.$Commercial implications: Allows photo sharing apps to offer enhanced location privacy features by embedding GeoEdit’s editing techniques.
- •For social media platform teams: Use GeoEdit to reduce risks of location doxing by applying subtle image edits on uploaded photos.
Authors
Yu Yan, Shengjia Cheng, Fei Ma, Zhe Wang, Peng Chen
Abstract
Recent efforts have demonstrated the strong capabilities of Multimodal Large Reasoning Models (MLRMs) in geographic reasoning. By jointly analyzing subtle visual cues, these models can infer the locations of photos, thus enabling doxing even by non-expert adversaries. Existing defense methods struggle to balance the shareability of photos with effective protection, and often fail to sufficiently disrupt the implicit geo-evidence chains embedded in scene semantics. To this end, we propose GeoEdit, an adversarial de-evidencing framework for silent geoprivacy protection via image editing. Specifically, we apply minimal edits to high-risk evidence anchors in the image, thereby disrupting the underlying spatial reasoning chains while preserving overall scene semantics. We identify these anchors by combining structured knowledge with adversarially discovered geographic cues. To further navigate the large editing action space, we adopt a learnable policy to iteratively refine editing decisions, distilling successful trajectories into transferable natural-language insights for consistent and efficient editing. Evaluated on LifePhotoBench, GeoEdit achieves strong protection against 6 mainstream MLRMs, outperforming visual perturbation baselines by over 0.59 in image fidelity.