A Comprehensive Security Evaluation Framework for Chinese Large Language Models
This paper presents a pioneering security evaluation framework for Chinese generative large language models (LLMs), addressing a significant gap in the field. The framework is supported by a comprehensive dataset of over 12,000 samples, systematically organized across four key dimensions: value evaluation, robustness, training data security, and prompt injection attacks. Each dimension is further subdivided into detailed subcategories.
Authors
- Zhaohong Jia (ORCID: https://orcid.org/0000-0001-6607-7025)
- Junfeng Fang (ORCID: https://orcid.org/0000-0002-3317-2103)
- Zhenhua Huang (ORCID: https://orcid.org/0000-0003-3178-9721)
- Yuli Yang (ORCID: https://orcid.org/0009-0001-6485-6583)
Institutions
- University of Science and Technology of China (CN)
- Anhui University (CN)
- National University of Singapore (SG)
Publication Details
- Published
- 2025-05-08
- DOI
- https://doi.org/10.1145/3701716.3717528
- Primary Topic
- Network Security and Intrusion Detection
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- National Natural Science Foundation of China