Black-Box Guardrail Reverse-engineering Attack

Large language models (LLMs) increasingly employ guardrails to enforce ethical, legal, and application-specific constraints on their outputs. While effective at mitigating harmful responses, these guardrails introduce a new class of vulnerabilities by exposing observable decision patterns. In this work, we present the first study of black-box LLM guardrail reverse-engineering attacks. We propose Guardrail Reverse-engineering Attack (GRA), a reinforcement learning-based framework that leverages genetic algorithm-driven data augmentation to approximate the decision-making policy of victim guardrails. By iteratively collecting input-output pairs, prioritizing divergence cases, and applying targeted mutations and crossovers, our method incrementally converges toward a high-fidelity surrogate of the victim guardrail. We evaluate GRA on three widely deployed commercial systems, namely ChatGPT, DeepSeek, and Qwen3, and demonstrate that it achieves an rule matching rate exceeding 0.92 while requiring less than $85 in API costs. These findings underscore the practical feasibility of guardrail extraction and highlight significant security risks for current LLM safety mechanisms. Our findings expose critical vulnerabilities in current guardrail designs and highlight the urgent need for more robust defense mechanisms in LLM deployment.

Key Contributions

• First systematic black-box guardrail reverse-engineering attack (GRA) using a reinforcement learning framework with genetic algorithm-driven data augmentation to clone victim guardrail policies
• A legal-moral evaluation dataset for benchmarking guardrail extraction fidelity and alignment performance
• Empirical demonstration on ChatGPT, DeepSeek, and Qwen3 showing >0.92 rule matching rate at under $85 API cost

🛡️ Threat Analysis

Details

Similar Papers