Desafios técnicos e computacionais na detecção de phishing em dispositivos móveis: revisão bibliográfica

Abstract

The rapid growth in mobile device usage over recent decades has reshaped communication and digital transactions while simultaneously creating a fertile ground for phishing attacks. This study presents a literature review on the technical and computational challenges involved in detecting phishing on mobile devices, with particular emphasis on smishing, the SMS-based variant, and on machine learning and deep learning approaches. Through the analysis of nineteen papers published between 2022 and 2026, the main algorithms employed were identified, along with reported accuracy results and recurring limitations such as dependence on restricted datasets, difficulty generalising across languages other than English, and the computational cost of more complex models. Findings indicate that classical algorithms and deep learning architectures achieve high performance in controlled settings but face significant obstacles when deployed in real-world scenarios on resource-constrained devices. It is concluded that progress in this field requires both the development of lighter and more adaptable solutions and the creation of multilingual, regularly updated datasets.