Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Defending Large Language Models Against Jailbreaking Attacks Through Goal Prioritization

About

While significant attention has been dedicated to exploiting weaknesses in LLMs through jailbreaking attacks, there remains a paucity of effort in defending against these attacks. We point out a pivotal factor contributing to the success of jailbreaks: the intrinsic conflict between the goals of being helpful and ensuring safety. Accordingly, we propose to integrate goal prioritization at both training and inference stages to counteract. Implementing goal prioritization during inference substantially diminishes the Attack Success Rate (ASR) of jailbreaking from 66.4% to 3.6% for ChatGPT. And integrating goal prioritization into model training reduces the ASR from 71.0% to 6.6% for Llama2-13B. Remarkably, even in scenarios where no jailbreaking samples are included during training, our approach slashes the ASR by half. Additionally, our findings reveal that while stronger LLMs face greater safety risks, they also possess a greater capacity to be steered towards defending against such attacks, both because of their stronger ability in instruction following. Our work thus contributes to the comprehension of jailbreaking attacks and defenses, and sheds light on the relationship between LLMs' capability and safety. Our code is available at \url{https://github.com/thu-coai/JailbreakDefense_GoalPriority}.

Zhexin Zhang, Junxiao Yang, Pei Ke, Fei Mi, Hongning Wang, Minlie Huang• 2023

Related benchmarks

TaskDatasetResultRank
Mathematical ReasoningGSM8K (test)
Accuracy99
954
Instruction FollowingAlpacaEval
Win Rate82
420
Multitask Language UnderstandingMMLU (test)
Accuracy85
312
Jailbreak DefenseJBB-Behaviors
ASR0.00e+0
121
Jailbreak DefenseAdvBench--
115
Jailbreak DefenseWild Jailbreak
ASR0.5
114
Jailbreak DefensePAIR
ASR6
97
Jailbreak DefenseGCG
ASR0.00e+0
91
Jailbreak DefenseHarmBench
PAIR ASR0.5
91
Instruction FollowingVicunaEval
Rouge-L35
72
Showing 10 of 47 rows

Other info

Code

Follow for update