Using Context and Interactions to Verify User-Intended Network Requests.
Shuang He, Michelle Y. Wong, David Lie · arXiv (Cornell University) · 2020
Client-side malware can attack users by tampering with applications or user interfaces to generate requests that users did not intend. We propose Verified Intention (VInt), which ensures a network request, as received by a service, is user-intended. VInt is based on seeing what the user (context). VInt screenshots the user interface as the user interacts with a security-sensitive form. There are two main components. First, VInt ensures output integrity and authenticity by validating the context, ensuring the user sees correctly rendered information. Second, VInt extracts user-intended inputs from the on-screen user-provided inputs, with the assumption that a human user checks what they entered. Using the user-intended inputs, VInt deems a request to be user-intended if the request is generated properly from the user-intended inputs while the user is shown the correct information. VInt is implemented using image analysis and Optical Character Recognition (OCR). Our evaluation shows that VInt is accurate and efficient.