PhishAR: Using Augmented Reality to help users make better security decisions

点击翻译按钮获取中文摘要

Computer users must repeatedly decide whether to trust a received email or to consider it a phishing attempt. Knowledgeable users develop various phishing detection mechanisms: they detect unexpected spelling mistakes, spot suspiciously urgent sentiment, or check for mismatches between the sender's email domain and their claimed identity.These nuances, while complex for non-expert users to learn, lend themselves to being automated by machine learning models. We are working on an augmented reality based security assistant: a mobile device system that uses computer vision and machine learning techniques together with immersive AR capabilities of modern smartphones and headsets to supervise and provide seamless security guidance to non-expert users by looking at the screens of other devices and instructing them on how to use them.More specifically, after an email is received, our solution allows the user to scan it with their mobile phone and receive an estimate of the likelihood of the email being a fraudulent phishing attempt, together with interactive explanations as to why such a conclusion was made.