RvB: Automating AI System Hardening via Iterative Red-Blue Games

TL;DR

This paper introduces RvB, a game-theoretic framework for automated, iterative AI system hardening that enhances robustness against vulnerabilities without requiring parameter updates.

Contribution

The paper presents a novel training-free, sequential game framework for dynamic AI security hardening, demonstrating its effectiveness across multiple challenging domains.

Findings

Achieves 90% defense success rate against code vulnerabilities

Attains 45% success rate in guardrail optimization

Maintains near 0% false positive rate

Abstract

The dual offensive and defensive utility of Large Language Models (LLMs) highlights a critical gap in AI security: the lack of unified frameworks for dynamic, iterative adversarial adaptation hardening. To bridge this gap, we propose the Red Team vs. Blue Team (RvB) framework, formulated as a training-free, sequential, imperfect-information game. In this process, the Red Team exposes vulnerabilities, driving the Blue Team to learning effective solutions without parameter updates. We validate our framework across two challenging domains: dynamic code hardening against CVEs and guardrail optimization against jailbreaks. Our empirical results show that this interaction compels the Blue Team to learn fundamental defensive principles, leading to robust remediations that are not merely overfitted to specific exploits. RvB achieves Defense Success Rates of 90\% and 45\% across the respective tasks while maintaining near 0\% False Positive Rates, significantly surpassing baselines. This work establishes the iterative adversarial interaction framework as a practical paradigm that automates the continuous hardening of AI systems.