Ctta: a novel chain-of-thought transfer adversarial attacks framework for large language models

Today's article comes from the journal of Cybersecurity. The authors are Yue et al., from Henan University of Science and Technology, in China. In this paper they explore how chain-of-thought prompting (CoT) might be exposing a new attack surface for bad-actors.

DOI: 10.1186/s42400-024-00338-1