Defending Large Language Models against Jailbreak Attacks via Semantic Smoothing

arXiv:2402.16192v1 Announce Type: new
Abstract: Aligned large language models (LLMs) are vulnerable to jailbreaking attacks, which bypass the safeguards of targeted LLMs and fool them into generating objectionable content. While initial defenses show promise against token-based threat models, there do not exist defenses that provide robustness against semantic attacks and avoid unfavorable trade-offs between robustness and nominal performance. To meet this need, we propose SEMANTICSMOOTH, a smoothing-based defense that aggregates the predictions of multiple semantically transformed copies of a given …

arxiv attacks cs.cl jailbreak language language models large language large language models semantic type via