الملخص
Safety-aligned language models often refuse a harmful request stated directly but answer the same request inside a role-play or narrative wrapper. We measure this vulnerability across languages and registers: attack success on Qwen3-1.7B is already 89.4% in English and 93.0% in modern Chinese, and reaches 95.7% in Classical Chinese. We build GUISE, a benchmark for systematically studying this vulnerability. It includes parallel requests in English, modern Chinese, and Classical Chinese, matched harmful and benign pairs, wrapper types held out for evaluation, and a stricter criterion that counts warn-then-answer responses as attack successes. Representation analysis shows that language and register move harmful-request representations only slightly away from the model's refusal direction, whereas narrative wrappers move them much farther away. We propose AXIS, which combines preference optimisation with a rotation objective that aligns harmful-request representations with the refusal direction and a commitment objective that trains the model to refuse completely rather than produce a warn-then-answer response. Across Qwen3-1.7B, Qwen3-4B and GLM-4-9B, AXIS achieves the highest combined safety and usability score among the compared methods.
الكلمات المفتاحية
بيانات النشر
- المجلة
- غير متاح
- وصول مفتوح
- وصول مفتوح أخضر
اقتبس هذه المقالة
APA 7
Ye, Z., Wang, Y., Jin, Y., Mei, B., Li, F., Wang, W., Gao, S., & Liu, X. (2026). How Narrative Wrapping Affects LLM Refusal: A Cross-Language Benchmark and Defense. https://omanscience.com/ar/articles/how-narrative-wrapping-affects-llm-refusal-a-cross-language-benchmark-and-defense
MLA 9
Ye, Zhankai, et al. "How Narrative Wrapping Affects LLM Refusal: A Cross-Language Benchmark and Defense." https://omanscience.com/ar/articles/how-narrative-wrapping-affects-llm-refusal-a-cross-language-benchmark-and-defense.
شيكاغو (المؤلف–التاريخ)
Ye, Zhankai, Yanning Wang, Yukai Jin, Bo Mei, Fangyi Li, Wei Wang, Shangqian Gao, and Xin Liu. 2026. "How Narrative Wrapping Affects LLM Refusal: A Cross-Language Benchmark and Defense." https://omanscience.com/ar/articles/how-narrative-wrapping-affects-llm-refusal-a-cross-language-benchmark-and-defense.
هارفارد
Ye, Z., Wang, Y., Jin, Y., Mei, B., Li, F., Wang, W., Gao, S. and Liu, X. (2026) 'How Narrative Wrapping Affects LLM Refusal: A Cross-Language Benchmark and Defense', Available at: https://omanscience.com/ar/articles/how-narrative-wrapping-affects-llm-refusal-a-cross-language-benchmark-and-defense.
فانكوفر
Ye Z, Wang Y, Jin Y, Mei B, Li F, Wang W, et al. How Narrative Wrapping Affects LLM Refusal: A Cross-Language Benchmark and Defense. https://omanscience.com/ar/articles/how-narrative-wrapping-affects-llm-refusal-a-cross-language-benchmark-and-defense
IEEE
Z. Ye, Y. Wang, Y. Jin, B. Mei, F. Li, W. Wang, S. Gao, and X. Liu, "How Narrative Wrapping Affects LLM Refusal: A Cross-Language Benchmark and Defense," https://omanscience.com/ar/articles/how-narrative-wrapping-affects-llm-refusal-a-cross-language-benchmark-and-defense.