There are some authors in common with a more recent paper "Defending Large Language Models against Jailbreak Attacks via Semantic Smoothing" https://arxiv.org/abs/2402.16192