An Overview of Criminal Exploitation and Abuse of Artificial Intelligence

The rapid advancement of Artificial Intelligence (AI) has led to its integration across various sectors of society. However, AI has also become increasingly embedded in criminal and harmful activities, thereby amplifying existing vulnerabilities and introducing new threats. This article examines pertinent literature, reports, and key incidents to develop a typology of the malicious use and abuse of AI-enabled systems. The primary goal is to categorize these activities and assess the associated risks. We begin by identifying the vulnerabilities inherent in AI models and discussing how malicious actors exploit these weaknesses. Following this, we explore AI-enabled and AI-enhanced attacks. While our review is comprehensive, it is not meant to be a definitive or exhaustive classification. Instead, we aim to provide an overview of the risks associated with advanced AI applications, thereby contributing to the ongoing discourse on this topic. Specifically, we identify four categories of AI abuse (integrity attacks, unintended AI outcomes, algorithmic trading, membership inference attacks) and four categories of AI misuse (socialengineering, misinformation/fake news, hacking, autonomous weapon systems). By mapping these threats, we facilitate a deeper consideration of governance strategies, policies, and actions that can be developed or refined to mitigate risks and prevent harmful outcomes. Strengthening collaboration among governments, industries, and civil society is essential to enhance preparedness and resilience against the malicious use and abuse of AI.