CenterYou++: Lightweight Privacy Enforcement for Android Using Hybrid AI and Cloud Decision Support

Seyedmostafa Safavi, Zarina Shukur, Muhammad Ehsan Rana
2025 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC), 2025
DOI: 10.1109/ASSIC64892.2025.11158601

Download PDF (accepted manuscript)Published version at IEEE Xplore

Abstract

While the Android operating system provides a vast application ecosystem offering extensive functionalities, it concurrently introduces substantial challenges to preserving user privacy. Mobile applications frequently engage in widespread collection of personal data, and users often encounter difficulties in managing application permissions effectively, creating exposure to potential privacy infringements. Existing techniques for privacy enforcement frequently encounter difficulties in achieving an optimal balance among scalability, user-centric design principles, and the critical need for timely responsiveness. To overcome these limitations, this work introduces CenterYou++, a novel privacy enforcement framework engineered specifically for Android environments. This framework employs a hybrid architecture, distinctively merging the capabilities of lightweight, on-device artificial intelligence (AI) agents with the resources of cloud-based decision support. The on-device AI facilitates low-latency, localized privacy enforcement actions grounded in learned behavioral patterns and predefined rules. Concurrently, the cloud component furnishes access to current privacy policies and global threat intelligence, while also underpinning the system's scalability. This synergistic design yields considerable advantages, encompassing real-time privacy enforcement capabilities, a more intuitive user experience achieved through automation and transparency, and strengthened policy enforcement mechanisms. The overarching objective is to establish an equilibrium between robust privacy protection and a fluid, unobtrusive user experience.

This is the authors' accepted manuscript. The final version is published by IEEE. © 2025 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.