World Journal of Engineering Research and Technology

E. Sree Shanmitha*, Kommineni Venkata Naga Hyma, Palle Venu Madhav and Banothu Ramji

ABSTRACT

The rapid evolution of Artificial Intelligence (AI) has impacted nearlyevery sector of society, with its integration into criminal activitiesposing new and expanded threats. This article reviews pertinentliterature, reports, and incidents to establish a typology of malicious AIuse and abuse. Our aim is to delineate various activities and associatedrisks, starting with an examination of AI model vulnerabilities and howthey can be exploited by malicious actors. Subsequently, we exploreboth AI-enabled and AI-enhanced attacks, providing a comprehensiveoverview without striving for exhaustive classification. We proposefour categories of malicious AI abuse: integrity attacks, unintended AIoutcomes, algorithmic trading, and membership inference attacks.Additionally, we identify four types of malicious AI use: social engineering, dissemination ofmisinformation/fake news, hacking, and the development of autonomous weapon systems. Bymapping these threats, we facilitate a deeper understanding of governance strategies, policies,and activities necessary to mitigate risks and prevent harm. Effective collaboration amonggovernments, industries, and civil society actors is crucial to enhance preparedness andresilience against the malicious use and abuse of AI. This article contributes to the expanding knowledge base on this issue and underscores the importance of proactive measures toaddress emerging threats in AI security.

[Full Text Article] [Download Certificate]