TrailBlazer: History-Guided Reinforcement Learning for Black-Box LLM Jailbreaking
This work addresses the critical safety issue of jailbreaking LLMs for adversarial research, offering a more efficient approach, though it is incremental as it builds on existing RL-based methods.
The paper tackled the problem of inefficient and unstable black-box LLM jailbreaking by proposing a history-aware reinforcement learning framework that analyzes and reweights vulnerability signals from prior interaction steps. The method achieved state-of-the-art jailbreak performance on AdvBench and HarmBench while significantly improving query efficiency.
Large Language Models (LLMs) have become integral to many domains, making their safety a critical priority. Prior jailbreaking research has explored diverse approaches, including prompt optimization, automated red teaming, obfuscation, and reinforcement learning (RL) based methods. However, most existing techniques fail to effectively leverage vulnerabilities revealed in earlier interaction turns, resulting in inefficient and unstable attacks. Since jailbreaking involves sequential interactions in which each response influences future actions, reinforcement learning provides a natural framework for this problem. Motivated by this, we propose a history-aware RL-based jailbreak framework that analyzes and reweights vulnerability signals from prior steps to guide future decisions. We show that incorporating historical information alone improves jailbreak success rates. Building on this insight, we introduce an attention-based reweighting mechanism that highlights critical vulnerabilities within the interaction history, enabling more efficient exploration with fewer queries. Extensive experiments on AdvBench and HarmBench demonstrate that our method achieves state-of-the-art jailbreak performance while significantly improving query efficiency. These results underscore the importance of historical vulnerability signals in reinforcement learning-driven jailbreak strategies and offer a principled pathway for advancing adversarial research on LLM safeguards.