🤖 AI Summary
Traditional cybersecurity approaches struggle to address the probabilistic nature of generative AI, rendering systems vulnerable to emerging threats such as model inversion, data poisoning, and prompt injection attacks. This work proposes the first STRIDE-based threat modeling framework tailored specifically for generative AI, which seamlessly integrates the NIST AI Risk Management Framework (RMF) with the OWASP LLM Top 10 vulnerability taxonomy. The framework establishes a comprehensive six-phase AI security assessment lifecycle and is accompanied by a dedicated web-based tool. Evaluated in a sandboxed LLM chatbot environment, the approach significantly reduces attack success rates from 80% to 15%, effectively bridging the gap between high-level risk governance and low-level technical defenses.
📝 Abstract
Traditional cybersecurity methodologies target deterministic systems and fail to address the probabilistic nature of AI, leaving systems vulnerable to attack vectors such as model inversion, data poisoning, and prompt injection. Recent industry reports indicate that a majority of organizations deploying AI lack a dedicated security strategy, with adversarial attacks increasing rapidly year-over-year. We present \textit{STRIDE-AI}, a framework that bridges the gap between high-level risk standards (NIST AI RMF) and technical vulnerability taxonomies (OWASP LLM Top 10). The framework defines a six-phase assessment lifecycle, introduces a threat modeling adaptation of classical STRIDE for AI systems, and is operationalized through a purpose-built web tool. We provide an initial validation of the approach through a black-box assessment of a deployed LLM chatbot, which successfully reduced the attack success rate from 80\% to 15\% in our sandbox case study.