Announcement_4
Sockpuppetting, my work on jailbreaking LLMs via prefilling and optimization, was accepted at ICML’s Second Workshop on Agents in the Wild: Safety, Security, and Beyond. Thank you to anyone who stopped by my poster! [paper]
Sockpuppetting, my work on jailbreaking LLMs via prefilling and optimization, was accepted at ICML’s Second Workshop on Agents in the Wild: Safety, Security, and Beyond. Thank you to anyone who stopped by my poster! [paper]