Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2601.00588 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908746402234368 |
|---|---|
| author | Zhou, Zhenhong Yan, Shilinlu Liu, Chuanpu Li, Qiankun Wang, Kun Zeng, Zhigang |
| author_facet | Zhou, Zhenhong Yan, Shilinlu Liu, Chuanpu Li, Qiankun Wang, Kun Zeng, Zhigang |
| contents | Large language models (LLMs) are increasingly deployed in cost-sensitive and on-device scenarios, and safety guardrails have advanced mainly in English. However, real-world Chinese malicious queries typically conceal intent via homophones, pinyin, symbol-based splitting, and other Chinese-specific patterns. These Chinese-specific adversarial patterns create the safety evaluation gap that is not well captured by existing benchmarks focused on English. This gap is particularly concerning for lightweight models, which may be more vulnerable to such specific adversarial perturbations. To bridge this gap, we introduce the Chinese-Specific Safety Benchmark (CSSBench) that emphasizes these adversarial patterns and evaluates the safety of lightweight LLMs in Chinese. Our benchmark covers six domains that are common in real Chinese scenarios, including illegal activities and compliance, privacy leakage, health and medical misinformation, fraud and hate, adult content, and public and political safety, and organizes queries into multiple task types. We evaluate a set of popular lightweight LLMs and measure over-refusal behavior to assess safety-induced performance degradation. Our results show that the Chinese-specific adversarial pattern is a critical challenge for lightweight LLMs. This benchmark offers a comprehensive evaluation of LLM safety in Chinese, assisting robust deployments in practice. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_00588 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CSSBench: Evaluating the Safety of Lightweight LLMs against Chinese-Specific Adversarial Patterns Zhou, Zhenhong Yan, Shilinlu Liu, Chuanpu Li, Qiankun Wang, Kun Zeng, Zhigang Computation and Language Large language models (LLMs) are increasingly deployed in cost-sensitive and on-device scenarios, and safety guardrails have advanced mainly in English. However, real-world Chinese malicious queries typically conceal intent via homophones, pinyin, symbol-based splitting, and other Chinese-specific patterns. These Chinese-specific adversarial patterns create the safety evaluation gap that is not well captured by existing benchmarks focused on English. This gap is particularly concerning for lightweight models, which may be more vulnerable to such specific adversarial perturbations. To bridge this gap, we introduce the Chinese-Specific Safety Benchmark (CSSBench) that emphasizes these adversarial patterns and evaluates the safety of lightweight LLMs in Chinese. Our benchmark covers six domains that are common in real Chinese scenarios, including illegal activities and compliance, privacy leakage, health and medical misinformation, fraud and hate, adult content, and public and political safety, and organizes queries into multiple task types. We evaluate a set of popular lightweight LLMs and measure over-refusal behavior to assess safety-induced performance degradation. Our results show that the Chinese-specific adversarial pattern is a critical challenge for lightweight LLMs. This benchmark offers a comprehensive evaluation of LLM safety in Chinese, assisting robust deployments in practice. |
| title | CSSBench: Evaluating the Safety of Lightweight LLMs against Chinese-Specific Adversarial Patterns |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2601.00588 |