AndroidLeaks: automatically detecting potential privacy leaks in android applications on a large scale

摘要: As mobile devices become more widespread and powerful, they store sensitive data, which includes not only users' personal information but also the data collected via sensors throughout day. When applications have access to this growing amount of information, may leak it carelessly or maliciously. Google's Android operating system provides a permissions-based security model that restricts an application's user's private data. Each application statically declares functionality requires in manifest, is presented user upon installation. However, clear how used once installed. To combat problem, we present AndroidLeaks, static analysis framework for automatically finding potential leaks on massive scale. AndroidLeaks drastically reduces number traces auditor has verify manually. We evaluate efficacy 24,350 from several markets. found 57,299 privacy 7,414 applications, out manually verified 2,342 including phone GPS location, WiFi audio recorded with microphone. examined these 30 hours, indicates capable scaling increasingly large set available applications.