Exploring Covert Third-party Identifiers through External Storage in the Android New Era

Third-party tracking plays a vital role in the mobile app ecosystem, which relies on identifiers to gather user data across multiple apps. In the early days of Android, tracking SDKs could effortlessly access non-resettable hardware identifiers for third-party tracking. However, as privacy concerns mounted, Google has progressively restricted device identifier usage through Android system updates. In the new era, tracking SDKs are only allowed to employ user-resettable identifiers which users can also opt out of, prompting SDKs to seek alternative methods for reliable user identification across apps. In this paper, we systematically explore the practice of third-party tracking SDKs covertly storing their own generated identifiers on external storage, thereby circumventing Android's identifier usage restriction and posing a considerable threat to user privacy. We devise an analysis pipeline for an extensive large-scale investigation of this phenomenon, leveraging kernel-level instrumentation and UI testing techniques to automate the recording of app file operations at runtime. Applying our pipeline to 8,000 Android apps, we identified 17 third-party tracking SDKs that store identifiers on external storage. Our analysis reveals that these SDKs employ a range of storage techniques, including hidden files and attaching to existing media files, to make their identifiers more discreet and persistent. We also found that most SDKs lack adequate security measures, compromising the confidentiality and integrity of identifiers and enabling deliberate attacks. Furthermore, we examined the impact of Scoped Storage -Android's latest defense mechanism for external storage on these covert third-party identifiers, and proposed a viable exploit that breaches such a defense mechanism. Our work underscores the need for greater scrutiny of third-party tracking practices and better solutions to safeguard user privacy in the Android ecosystem.