Newest papers

Not Where You Think: Protecting Location Privacy Against Reasoning-Based Geolocation

Shanghao Shi, Jian Xiao, Hao Li, Yuanhaur Chang, Yevgeniy Vorobeychik, Nathan Jacobs, Neal Patwari, Ning Zhang · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.30439 · cited by 0 works (OpenAlex) · open-access PDF

Russian

Чтобы понять механизм, лежащий за такой мелкозернистой локализацией, мы разрабатываем GeoTrace, структурированную аргументационную атаку, которая раскрывает, как модели извлекают визуальные подсказки, относящиеся к местоположению, и составляют их через грубое географическое обоснование, даже когда удаляются метаданные и очевидные географические идентификаторы эффективного местоположения. Мы оцениваем GeoTrace на мировом целевом показателе 525 обычных изображений, охватывающих природные, городские, жилые и центрированные на человека сцены, исключая знаменитые достопримечательности и индикаторы прямого местоположения. Наши результаты показывают, что GeoTrace локализует 25.71% изображений в пределах 1 км с Gemini-3

Machine translation (Meta M2M100); the English original follows.

Abstract

Recent large reasoning models can infer where ordinary photos were taken using only visual information, creating a serious location privacy risk even when metadata and obvious geographic identifiers are removed. To understand the mechanism behind such fine-grained localization, we develop GeoTrace, a structured reasoning attack that exposes how models extract location-relevant visual clues and compose them through coarse-to-fine geographic reasoning. We evaluate GeoTrace on a worldwide benchmark of 525 ordinary images spanning natural, urban, residential, and human-centered scenes while excluding famous landmarks and direct location indicators. Our results show that GeoTrace localizes 25.71% of images within 1 km with Gemini-3-Flash and 16.95% with GPT-5-mini. More importantly, our analysis reveals a key insight in the visual clue chain: broad scene properties mainly support coarse localization, whereas precise localization disproportionately relies on small, distinctive, and editable local clues. Building on this insight, we develop GeoShift, a proactive user-side defense that breaks the visual clue chain before image release. Rather than simply removing geographic evidence, GeoShift preserves stable scene properties, selects a visually compatible false target location, and strategically rewrites high-impact geo-revealing clues to construct a coherent alternative geographic hypothesis. Experiments show that GeoShift reduces street-level localization from 16.95% to 2.48% under GPT-5-mini and from 25.71% to 3.81% under Gemini-3-Flash, while preserving visual quality and remaining effective against adaptive attacks.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Geolocation, Computer science, Metadata, Artificial intelligence, Computer vision, Construct (python library)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange