Browser History Stealing with Captive Wi-Fi Portals

Abstract

In this paper we show that HSTS headers and long-term cookies (like those used for user tracking) are so prevailing that they allow a malicious Wi-Fi operator to gain significant knowledge about the past browsing history of users. We demonstrate how to combine both into a history stealing attack by including specially crafted references into a captive portal or by injecting them into legitimate HTTP traffic. Captive portals are used on many Wi-Fi Internet hotspots to display the user a message, like a login page or an acceptable use policy before they are connected to the Internet. They are typically found in public places such as airports, train stations, or restaurants. Such systems have been known to be troublesome for many reasons. In this paper we show how a malicious operator can not only gain knowledge about the current Internet session, but also about the user's past. By invisibly placing vast amounts of specially crafted references into these portal pages, we can lure the browser into revealing a user's browsing history by either reading stored persistent (long-term) cookies or evaluating responses for previously set HSTS headers. An occurrence of a persistent cookie, as well as a direct call to the pages' HTTPS site is a reliable sign of the user having visited this site earlier. Thus, this technique allows for a site-based history stealing, similar to the famous link-color history attacks. For the Alexa Top 1,000 sites, between 82% and 92% of sites are effected as they use persistent cookies over HTTP. For the Alexa Top 200,000 we determined the number of vulnerable sites between 59% and 86%. We extended our implementation of this attack by other privacy-invading attacks that enrich the collected data with additional personal information.

DOI: 10.1109/SPW.2016.42

4 Figures and Tables

Cite this paper

@article{Dabrowski2016BrowserHS, title={Browser History Stealing with Captive Wi-Fi Portals}, author={Adrian Dabrowski and Georg Merzdovnik and Nikolaus Kommenda and Edgar R. Weippl}, journal={2016 IEEE Security and Privacy Workshops (SPW)}, year={2016}, pages={234-240} }