AN OUTLINE OF AI AND SOCIAL CYBERSECURITY: TECHNIQUES FOR SPOTTING THREATS, EVALUATIONS, CHALLENGES, AND POSSIBLE

Abstract

Due to the rapid expansion of several online and digital cyberthreats, social cybersecurity has emerged as one of the most important areas of research for contemporary digital communication systems. Artificial intelligence (AI) is being used more and more by both attackers and defenders, which has significantly changed cybersecurity. The most plausible forms of cybercriminals' malicious communication, such as phishing attacks, misinformation campaigns, fake social media accounts, the use of deepfakes, spam content, and automated social engineering attacks, are now automated by cybercriminals using more recent techniques like artificial intelligence (AI) for both fraud and AI attacks and tactics. Deepfakes and other sophisticated AI algorithms were used to introduce the threat of social media. In order to detect and prevent these threats more quickly, researchers and companies have started integrating artificial intelligence into defensive systems. In order to address this, a survey paper is carried out to investigate social cybersecurity and artificial intelligence from the perspective of machine learning, including contemporary machine learning-based attack detection techniques, methods for evaluating them, current issues, and potential future developments. Natural language processing (NLP), neural networks, supervised and unsupervised learning, transformer-based models like BERT, and other machine learning and deep learning approaches for cyber threat identification are also studied in this work. The study also examines AI applications, including deepfake recognition, phishing detection, bot identification, cyberbullying detection, and fake news detection. In order to describe AI-based detection systems, the study also assesses commonly used performance metrics including accuracy, precision, recall.