Security Risks of Agentic Vehicles: A Systematic Analysis of Cognitive and Cross-Layer Threats

2025-12-18

AI Summary

This paper addresses novel safety threats introduced by Agentic AI in Autonomous Guided Vehicles (AgVs), systematically uncovering cognitive-layer vulnerabilities and cross-layer interactions—spanning perception, control, and communication—that induce policy misjudgment and loss of vehicle control. Method: We propose the first structured safety risk analysis framework tailored for automotive-grade AgV platforms, innovatively incorporating a role-based hierarchical architecture comprising individual agents and driving-policy agents. The framework integrates role-oriented modeling, attack-chain analysis, cross-layer threat mapping, and severity matrix evaluation. Contribution/Results: We identify multiple previously undocumented cross-layer attack vectors not covered by OWASP, quantitatively characterize the propagation mechanism from minor perturbations to safety-critical failures, and establish a scalable, reproducible risk assessment benchmark for SAE Level 2+ to Level 4 autonomous driving systems.

Technology Category

• Multiagent Systems: Adversarial Agents
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Security and Privacy: Large-scale security measurements

Abstract

Agentic AI is increasingly being explored and introduced in both manually driven and autonomous vehicles, leading to the notion of Agentic Vehicles (AgVs), with capabilities such as memory-based personalization, goal interpretation, strategic reasoning, and tool-mediated assistance. While frameworks such as the OWASP Agentic AI Security Risks highlight vulnerabilities in reasoning-driven AI systems, they are not designed for safety-critical cyber-physical platforms such as vehicles, nor do they account for interactions with other layers such as perception, communication, and control layers. This paper investigates security threats in AgVs, including OWASP-style risks and cyber-attacks from other layers affecting the agentic layer. By introducing a role-based architecture for agentic vehicles, consisting of a Personal Agent and a Driving Strategy Agent, we will investigate vulnerabilities in both agentic AI layer and cross-layer risks, including risks originating from upstream layers (e.g., perception layer, control layer, etc.). A severity matrix and attack-chain analysis illustrate how small distortions can escalate into misaligned or unsafe behavior in both human-driven and autonomous vehicles. The resulting framework provides the first structured foundation for analyzing security risks of agentic AI in both current and emerging vehicle platforms.

Problem

• Analyzing security threats in agentic vehicles
• Investigating vulnerabilities in agentic AI and cross-layer risks
• Providing a framework for agentic AI security in vehicles

Innovation

• Introduces role-based agent architecture for vehicles
• Analyzes cross-layer security risks in agentic vehicles
• Provides severity matrix for attack-chain analysis