2

Apple iOS is one of the most popular mobile operating sys-tems. As its core security technology, iOS provides applica-tion sandboxing but assigns a generic sandboxing profile to every third-party application. However, recent attacks and incidents with benign applications demonstrate that this de-sign decision is vulnerable to crucial privacy and security breaches, allowing applications (either benign or malicious) to access contacts, photos, and device IDs. Moreover, the dynamic character of iOS apps written in Objective-C ren-ders the currently proposed static analysis tools less useful. In this paper, we aim to address the open problem of preventing (not only detecting) privacy leaks and simulta-neously strengthening security against runtime attacks on iOS. Compared to similar research work on the open An-droid, realizing such a system for the closed-source iOS is highly involved. We present the design and implementation of PSiOS, a tool that features a novel policy enforcement framework for iOS. It provides fine-grained, application-specific, and user/administrator defined sandboxing for each third-party application without requiring access to the application source code. Our reference implementation deploys control-flow integrity based on the recently proposed MoCFI (Mobile CFI) framework that only protects applications against run-time attacks. We evaluated several popular iOS applications (e.g., Facebook, WhatsApp) to demonstrate the efficiency and effectiveness of PSiOS