paper-with-me

Papers

Automatic Pseudo-Harmful Prompt Generation for Evaluating False Refusals in Large Language Models

2024-09-01 · Bang An, Sicheng Zhu, Ruiyi Zhang, Michael-Andrei Panaitescu-Liess, Yuancheng Xu, Furong Huang

Safety-aligned large language models (LLMs) sometimes falsely refuse pseudo-harmful prompts, like "how to kill a mosquito," which are actually harmless. Frequent false refusals not only frustrate users but also provoke a public backlash against the very values alignment seeks to protect. In this paper, we propose the first method to auto-generate diverse, content-controlled, and model-dependent pseudo-harmful prompts. Using this method, we construct an evaluation dataset called PHTest, which is ten times larger than existing datasets, covers more false refusal patterns, and separately labels controversial prompts. We evaluate 20 LLMs on PHTest, uncovering new insights due to its scale and labeling. Our findings reveal a trade-off between minimizing false refusals and improving safety against jailbreak attacks. Moreover, we show that many jailbreak defenses significantly increase the false refusal rates, thereby undermining usability. Our method and dataset can help developers evaluate and fine-tune safer and more usable LLMs. Our code and dataset are available at https://github.com/umd-huang-lab/FalseRefusal

📄 PDF Abstract BibTeX arXiv:2409.00598

Code (1)

umd-huang-lab/falserefusal 공식 구현

Similar Papers 제목 키워드 기반

Bayesian Evaluation of Large Language Model Behavior

2025-11-04 · Rachel Longjohn, Shang Wu, Saatvik Kher, Catarina Belém 외 arxiv

It is increasingly important to evaluate how text generation systems based on large language models (LLMs) behave, such as their tendency to produce harmful output or their sensitivity to adversarial inputs. Such evaluat…

Text Generation

TurboFuzzLLM: Turbocharging Mutation-based Fuzzing for Effectively Jailbreaking Large Language Models in Practice

2025-02-21 · Aman Goel, Xian Carrie Wu, Zhe Wang, Dmitriy Bespalov 외

Jailbreaking large-language models (LLMs) involves testing their robustness against adversarial prompts and evaluating their ability to withstand prompt attacks that could elicit unauthorized or malicious responses. In t…

HarmfulSkillBench: How Do Harmful Skills Weaponize Your Agents?

2026-04-16 · Yukun Jiang, Yage Zhang, Michael Backes, Xinyue Shen 외 arxiv

Large language models (LLMs) have evolved into autonomous agents that rely on open skill ecosystems (e.g., ClawHub and Skills.Rest), hosting numerous publicly reusable skills. Existing security research on these ecosyste…

AdvPrompter: Fast Adaptive Adversarial Prompting for LLMs

2024-04-21 · Anselm Paulus, Arman Zharmagambetov, Chuan Guo, Brandon Amos 외

Large Language Models (LLMs) are vulnerable to jailbreaking attacks that lead to generation of inappropriate or harmful content. Manual red-teaming requires a time-consuming search for adversarial prompts, whereas automa…

MMLURed Teaming

Be a Multitude to Itself: A Prompt Evolution Framework for Red Teaming

2025-02-22 · Rui Li, Peiyi Wang, Jingyuan Ma, Di Zhang 외

Large Language Models (LLMs) have gained increasing attention for their remarkable capacity, alongside concerns about safety arising from their potential to produce harmful content. Red teaming aims to find prompts that …

DiversityIn-Context LearningRed Teaming