Android accessibility features include a robust set of tools allowing developers to create apps for assisting people with disabilities. Unfortunately, this useful set of tools can also be abused and turned into an attack vector, providing malware with the ability to interact and read content from third-party apps. In this work, we are the first to study the impact that the stealthy exploitation of Android accessibility services can have on significantly reducing the forensic footprint of malware attacks, thus hindering both live and post-incident forensic investigations. We show that through Living off the Land (LotL) tactics, or by offering a malware-only substitute for attacks typically requiring more elaborate schemes, accessibility-based malware can be rendered virtually undetectable. In the LotL approach, we demonstrate accessibility-enabled SMS and command and control (C2) capabilities. As for the latter, we show a complete cryptocurrency wallet theft, whereby the accessibility trojan can hijack the entire withdrawal process of a widely used app, including two-factor authentication (2FA). In both cases, we demonstrate how the attacks result in significantly diminished forensic evidence when compared to similar attacks not employing accessibility tools, even to the extent of maintaining device take-over without requiring malware persistence.

The security of financial apps on smartphones is threatened by a class of advanced and persistent malware that can bypass all existing security measures. Strong cryptography and trusted onchip hardware modules are powerless against sophisticated attacks that supplant device owners through device input record/replay functionality, effectively hijacking their credentials, privileges, and actions. In this paper, we introduce Proof-of-Presence and Locality (PoPL), a new security measure that tackles this threat by leveraging sensors to prove the physical presence of device owners and therefore discriminate between malware-initiated transaction requests and legitimate ones. Moreover, PoPL neither imposes the expense of additional hardware nor compromises app usability. In order to demonstrate PoPL's practicality, we developed PoPLar, a challenge puzzle implementation of the PoPL concept that ensures usability even on limited screen sizes by the use of a dendrogram. We have made it available as an open-source library ready to be integrated with minimal effort with existing apps. We demonstrate PoPLar's effectiveness and ease of integration through case studies involving apps from the three top cryptocurrency exchanges and an open-source crypto wallet.

Sophisticated Android malware families often implement techniques aimed at avoiding detection. Split personality malware for example, behaves benignly when it detects that it is running on an analysis environment such as a malware sandbox, and maliciously when running on a real user's device. These kind of techniques are problematic for malware analysts, often rendering them unable to detect or understand the malicious behaviour. This is where sandbox hardening comes into play. In our work, we exploit sandbox detecting heuristic prediction to predict and automatically generate bytecode patches, in order to disable the malware's ability to detect a malware sandbox. Through the development of AndroNeo, we demonstrate the feasibility of our approach by showing that the heuristic prediction basis is a solid starting point to build upon, and demonstrating that when heuristic prediction is followed by bytecode patch generation, split personality can be defeated.