A global cyber-threat intelligence system with artificial intelligence and convolutional neural network

Abstract

Global cyber-attacks significantly impact the economy, society, organizations, and individuals. Existing research on cyber-attacks lacks in demonstrating Artificial Intelligence (AI) based analytical solutions for providing country-wide cyber threat intelligence. Cyber strategists at a national level require AI-based decision support systems for deciding a country’s cyber posture or preparedness. This paper proposes an AI-based solution that autonomously collects multidimensional cyber-attack data on social media posts on cyber-related outcry. The proposed system provides critical analytical capability in the cyber-threat spectrum and uses sophisticated AI-based algorithms for anomaly detection, prediction, sentiment analysis, location detection, translation, etc. The proposed system was deployed from 11 October 2022 to 31 October 2022. During these 21 days, the system autonomously collected 30,203 records on cyber threats covering multiple dimensions of cyber-threat. These dimensions included country-wide daily cyber-attack records by ransomware, exploits, web threats, spam, malicious mail, network attacks, local infections, and on-demand scan. Moreover, the system performed AI-based acquisition and analysis of 3789 cyber-related tweets from 3402 tweet users in 37 different languages. The system also autonomously translated 893 non-English tweets. The proposed system is the first solution that uses Convolutional Neural Network (CNN) based anomaly detection to detect abnormalities in cyber-threat spectrum worldwide and predict cyber-attacks automatically. The proposed system was demonstrated to provide evidence-based decisions on global cyber threats in multiple platforms, including iOS, Android, and Windows.