Universal Adversarial Triggers Are Not Universal

arXiv:2404.16020v1 Announce Type: new
Abstract: Recent work has developed optimization procedures to find token sequences, called adversarial triggers, which can elicit unsafe responses from aligned language models. These triggers are believed to be universally transferable, i.e., a trigger optimized on one model can jailbreak other models. In this paper, we concretely show that such adversarial triggers are not universal. We extensively investigate trigger transfer amongst 13 open models and observe inconsistent transfer. Our experiments further reveal a significant difference in …

abstract adversarial arxiv cs.cl jailbreak language language models one model optimization paper responses show token type universal work