ShieldLearner: A New Paradigm for Jailbreak Attack Defense in LLMs
This addresses the vulnerability of LLMs to evolving jailbreak attacks, offering a practical and efficient defense solution for real-world applications, though it appears incremental in improving adaptability and interpretability over existing methods.
The paper tackles the problem of defending LLMs against adversarial jailbreak attacks by proposing ShieldLearner, a novel paradigm that mimics human learning to autonomously distill attack signatures and synthesize defense heuristics, achieving a significantly higher defense success rate than existing baselines on both conventional and hard test sets with lower computational overhead.
Large Language Models (LLMs) have achieved remarkable success in various domains but remain vulnerable to adversarial jailbreak attacks. Existing prompt-defense strategies, including parameter-modifying and parameter-free approaches, face limitations in adaptability, interpretability, and customization, constraining their effectiveness against evolving threats. To address these challenges, we propose ShieldLearner, a novel paradigm that mimics human learning in defense. Through trial and error, it autonomously distills attack signatures into a Pattern Atlas and synthesizes defense heuristics into a Meta-analysis Framework, enabling systematic and interpretable threat detection. Furthermore, we introduce Adaptive Adversarial Augmentation to generate adversarial variations of successfully defended prompts, enabling continuous self-improvement without model retraining. In addition to standard benchmarks, we create a hard test set by curating adversarial prompts from the Wildjailbreak dataset, emphasizing more concealed malicious intent. Experimental results show that ShieldLearner achieves a significantly higher defense success rate than existing baselines on both conventional and hard test sets, while also operating with lower computational overhead, making it a practical and efficient solution for real-world adversarial defense.