Asset-Centric Threat Modeling for AI-Based Systems

Threat modeling for systems relying on Artificial In-telligence is not well explored. While conventional threat modeling methods and tools do not address AI-related threats, research on this amalgamation still lacks solutions capable of guiding and automating the process, as well as providing evidence that the methods hold up in practice. Consequently, this paper presents ThreatFinderAI, an approach and tool providing guidance and automation to model AI-related assets, threats, countermeasures, and quantify residual risks. To do so, ThreatFinderAI presents a novel AI-based stencil library for automated asset extraction, a threat knowledge graph spanning several community initiatives, and a novel method to identify business impacts of AI threats and an approach to quantify them. To evaluate the practicality of the approach, participants were tasked to recreate a threat model developed by cybersecurity experts of an AI-based healthcare platform. Secondly, the approach was used to identify and discuss strategic risks in an LLM-based application through a case study. Overall, the solution's usability was well-perceived and effectively supports threat identification and risk discussion.

I. Introduction

Artificial Intelligence (AI) is considered a disruptive tech-nology that is being integrated into different domains, ranging from healthcare to embedded implementations [1], which now serve as a key contributor to other technologies such as 6G [2]. In addition to the wide range of domains that show interest in AI technologies, an outstanding observation is the speed at which AI technology is adopted. For example, ChatGPT has attracted 100 million monthly active users within weeks [3].