ChemSafetyBench: Benchmarking LLM Safety on Chemistry Domain

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Haochen, Tang, Xiangru, Yang, Ziran, Han, Xiao, Feng, Xuanzhi, Fan, Yueqing, Cheng, Senhao, Jin, Di, Zhao, Yilun, Cohan, Arman, Gerstein, Mark
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910715902689280
author Zhao, Haochen
Tang, Xiangru
Yang, Ziran
Han, Xiao
Feng, Xuanzhi
Fan, Yueqing
Cheng, Senhao
Jin, Di
Zhao, Yilun
Cohan, Arman
Gerstein, Mark
author_facet Zhao, Haochen
Tang, Xiangru
Yang, Ziran
Han, Xiao
Feng, Xuanzhi
Fan, Yueqing
Cheng, Senhao
Jin, Di
Zhao, Yilun
Cohan, Arman
Gerstein, Mark
contents The advancement and extensive application of large language models (LLMs) have been remarkable, including their use in scientific research assistance. However, these models often generate scientifically incorrect or unsafe responses, and in some cases, they may encourage users to engage in dangerous behavior. To address this issue in the field of chemistry, we introduce ChemSafetyBench, a benchmark designed to evaluate the accuracy and safety of LLM responses. ChemSafetyBench encompasses three key tasks: querying chemical properties, assessing the legality of chemical uses, and describing synthesis methods, each requiring increasingly deeper chemical knowledge. Our dataset has more than 30K samples across various chemical materials. We incorporate handcrafted templates and advanced jailbreaking scenarios to enhance task diversity. Our automated evaluation framework thoroughly assesses the safety, accuracy, and appropriateness of LLM responses. Extensive experiments with state-of-the-art LLMs reveal notable strengths and critical vulnerabilities, underscoring the need for robust safety measures. ChemSafetyBench aims to be a pivotal tool in developing safer AI technologies in chemistry. Our code and dataset are available at https://github.com/HaochenZhao/SafeAgent4Chem. Warning: this paper contains discussions on the synthesis of controlled chemicals using AI models.
format Preprint
id arxiv_https___arxiv_org_abs_2411_16736
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle ChemSafetyBench: Benchmarking LLM Safety on Chemistry Domain
Zhao, Haochen
Tang, Xiangru
Yang, Ziran
Han, Xiao
Feng, Xuanzhi
Fan, Yueqing
Cheng, Senhao
Jin, Di
Zhao, Yilun
Cohan, Arman
Gerstein, Mark
Computation and Language
Artificial Intelligence
Chemical Physics
The advancement and extensive application of large language models (LLMs) have been remarkable, including their use in scientific research assistance. However, these models often generate scientifically incorrect or unsafe responses, and in some cases, they may encourage users to engage in dangerous behavior. To address this issue in the field of chemistry, we introduce ChemSafetyBench, a benchmark designed to evaluate the accuracy and safety of LLM responses. ChemSafetyBench encompasses three key tasks: querying chemical properties, assessing the legality of chemical uses, and describing synthesis methods, each requiring increasingly deeper chemical knowledge. Our dataset has more than 30K samples across various chemical materials. We incorporate handcrafted templates and advanced jailbreaking scenarios to enhance task diversity. Our automated evaluation framework thoroughly assesses the safety, accuracy, and appropriateness of LLM responses. Extensive experiments with state-of-the-art LLMs reveal notable strengths and critical vulnerabilities, underscoring the need for robust safety measures. ChemSafetyBench aims to be a pivotal tool in developing safer AI technologies in chemistry. Our code and dataset are available at https://github.com/HaochenZhao/SafeAgent4Chem. Warning: this paper contains discussions on the synthesis of controlled chemicals using AI models.
title ChemSafetyBench: Benchmarking LLM Safety on Chemistry Domain
topic Computation and Language
Artificial Intelligence
Chemical Physics
url https://arxiv.org/abs/2411.16736