Transparent Privacy Control via Static Information Flow Analysis

A common problem faced by modern mobile-device platforms is that third-party applications in the marketplace may leak private information without notifying users. Existing approaches adopted by these platforms provide little information on what applications will do with the private information, failing to effectively assist users in deciding whether to install applications and in controlling their privacy. To address this problem, we propose a transparent privacy control approach, where an automatic static analysis reveals to the user how private information is used inside an application. This flow information provides users with better insights, enabling them to determine when to use anonymized instead of real information, or to force script termination when scripts access private information. To further reduce the user burden in controlling privacy, our approach provides a default setting based on an extended information flow analysis that tracks whether private information is obscured before escaping through output channels. We built our approach into TouchDevelop, a novel application-creation environment that allows users to write application scripts on mobile devices, share them in a web bazaar, and install scripts published by other users. To evaluate our approach, we plan to study a portion of published scripts in order to evaluate the effectiveness and performance of information flow analysis. We also plan to carry out a user survey to evaluate the usability of our privacy control and guide our future design.