Implementación de un sistema basado en técnicas de machine learning para la detección de correos electrónicos de spear phishing

Abstract

The purpose of this project was to develop an automated solution capable of identifying malicious emails using large-scale language models (LLM) and machine learning techniques. For its implementation, an application programming interface (API) integrated into the n8n platform was designed, allowing for the establishment of a functional workflow for analyzing suspicious emails and automatically generating reports. Additionally, the Hybrid Analysis tool was incorporated to scan potentially malicious files and links, optimizing threat detection and classification processes. During the research, studies and official data were collected that demonstrate the increase in phishing and spear phishing attacks in Colombia, especially during the COVID-19 pandemic, highlighting the need to implement technological solutions based on artificial intelligence. The system was validated through controlled tests with simulated emails, yielding positive results in identifying and classifying fraudulent messages, demonstrating the effectiveness of the developed model.