Speak Easy: Eliciting Harmful Jailbreaks from LLMs with Simple Interactions.

Yik Siu Chan, Narutatsu Ri, Yuxin Xiao, Marzyeh Ghassemi

Browse the full ICML paper archive.