SOCIAL CYBERSECURITY AND ARTIFICIAL INTELLIGENCE: A SURVEY

Abstract

Over the past years, social cybersecurity has become one of the crucial research fields of modern digital communication systems because of the accelerated growth of various digital and web-based cyber threats. Both attackers and defenders have adopted ever-increasing use of artificial intelligence: AI, which has drastically transformed cybersecurity. Today, with the addition of newer methods like artificial intelligence (AI) for both fraud and AI attacks and tactics, cybercriminals use it to automate the most-explicable types of cyber criminals’ malicious communication, including phishing attacks, misinformation campaigns, fake social media accounts, the use of deepfakes, spam content and automated social engineering attacks. The threat of social media was launched via advanced AI algorithms such as deepfakes and so on. Consequently, researchers and organizations have begun incorporating artificial intelligence into defensive solutions for quicker detection and prevention of these threats. To answer this, a survey paper is conducted to explore social cybersecurity and artificial intelligence from the point of view of machine learning, such as modern machine learning-based attack detection techniques, evaluation methods for them as well as the issues of contemporary times and possibilities for the future. This work studies machine learning and deep learning techniques for cyber threat detection such as supervised learning, unsupervised learning, neural networks, natural language processing (NLP) and transformer-based models like BERT. The sections of this paper are organized into several sections. Here the literature review looks back at the previous works on AI driven cybersecurity systems and social cyber threat monitoring. The "Materials and methods" section gives insight of widely used datasets, algorithms, and evaluation metrics. The results and discussion section compares and analyzes advantages and limitations of various AI approaches. Last but not least, the paper identifies its current problems, future possibilities, and conclusions about the future of social cybersecurity.