Cybercrime in the AI Era: Definitions, Classification, Severity Assessment and the Role of AI in Combating Threats

Abstract

The growing prevalence, complexity and financial impact of cybercrimes pose significant challenges for law enforcement agencies worldwide. Many organizations are utilizing different technologies such as cloud technology to improve speed, accuracy, and reliability. However, without proper security measures, the risk is still vital against cyberattacks. Cybercrime can lead to various negative outcomes, including theft, fraud, financial losses, a decline in customer trust, and emotional consequences like fear, anger, and insecurity. This research aims to explore multiple definitions of cybercrime and identify the most effective categories for assessing its severity, focusing on key characteristics found in cybercrime descriptions. It also examines regulations designed to combat and prevent cybercrime, which Saudi Arabia and the UK are considering as Case studies. Additionally, the study explores the role of artificial intelligence, machine learning, deep learning, transformer models, and generative models in fighting cybercrime, especially their application in classification tasks. The research evaluates datasets used in previous studies, highlighting their features and providing insights into the nature and trends of cybercrime. The findings demonstrate how transformer models and generative AI approaches, such as BERT and GPT, have driven significant advancements in natural language processing tasks, improving cybercrime classification and severity assessment. Furthermore, the review underscores the importance of detailed datasets with case descriptions, demographic information, and clear labels, offering valuable insights into prevalent cybercrime methods and trends. It offers actionable recommendations for future research, emphasizing the need for interdisciplinary collaboration, robust datasets, and innovative AI approaches to address the evolving landscape of cybercrime.