Abstract

QR codes offer convenience by turning activities such as payments, sign-ins, ticket entries and menu browsing into one-step actions, however, their final destination may never reveal unless one scans them. It is quishing here that takes place by means of redirecting with the aid of QR code. This paper presents an overview of research findings related to phishing attacks and quishing using QR codes mainly published from 2023 – 2026. Machine-learning (ML), deep-learning (DL) and Explainable AI (XAI) techniques are all considered. Ten different publications have been considered with respect to the source of data, their detection methodologies, and the kind of explanations generated. The analysis reveals that there are not many cases where one model is developed which incorporates both detection of malicious QR images and a system of explainable URL analysis. The other main practical issues are that a warning coming just in time may not work because one has already fallen into a trap; reliance on predefined threats lists; non-interpretable predictions; legitimate websites wrongly flagged. This paper describes a theoretical framework which can merge information of QR codes and URL and calculate the threat of each case from 0 to 100 with the support of either SHAP or LIME for generating the explanation. This paper is just a conceptual proposal for further research in this field. It has neither been implemented nor is it based on any experimental evidence.

Keywords

How to Cite This Article

Rights