Metraipolis › archive › dc_7bdf6eb0043f
The "Knowledge-Behavior Gap" in Cultural Taboo Safety of Large Language Models — arXiv:2608.12341
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.12341
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-19T04:24:24.172Z
- SHA-256 of the archived text
8870b034f563f5ee4254a93f9102745405f83ae2f580d3d05e88038604082e64- Stable URL of this page
https://metraipolis.com/documento/dc_7bdf6eb0043f
How to cite
The "Knowledge-Behavior Gap" in Cultural Taboo Safety of Large Language Models — arXiv:2608.12341 — arXiv (en), https://arxiv.org/abs/2608.12341 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7bdf6eb0043f (Metraipolis archive dc_7bdf6eb0043f, sha256 8870b034f563f5ee4254a93f9102745405f83ae2f580d3d05e88038604082e64)
BibTeX
@misc{dc7bdf6eb0043f,
title = {The "Knowledge-Behavior Gap" in Cultural Taboo Safety of Large Language Models — arXiv:2608.12341},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.12341},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7bdf6eb0043f; sha256 8870b034f563f5ee4254a93f9102745405f83ae2f580d3d05e88038604082e64},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.12341. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Cultural taboo safety is essential for deploying large language models (LLMs), as culturally insensitive outputs may cause offense or even social harm. However, existing cultural benchmarks primarily assess cultural knowledge or values biases, while overlooking whether LLMs can recognize and respect cultural taboos, especially when taboos are implicitly hidden in seemingly harmless questions. Besides, cultural taboos are implicit, and context-dependent, thus poss unique challenges for reliable evaluation. To address these gaps, we introduce \textbf{CulShield}, the first public benchmark dedicated to evaluating and improving the cultural taboo safety of LLMs. CulShield spans 77 countries and territories, and includes over 2,020 taboos. It evaluates models along both explicit knowledge and implicit behaviors. Experiments on several advanced LLMs (e.g., GPT-4o-mini, Gemini-2.5-pro) reveal a clear ``knowledge-behavior gap'': models often fail to apply known taboos during interaction. We further show that variations in linguistic context can significantly affect LLMs' cultural taboo safety. Code and data is accessible here: this https URL .
Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-14 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.12341 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-19T04:17:42.369Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-14?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).