---
title: "低资源语言中的跨语言安全对齐幻象"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.11146v1"
permalink: "https://feed.astralor.com/items/itm_8ecaca672ff564d9de75e4c8"
published: "2026-08-11T17:05:26.000Z"
score: 77
channel: "ai"
tags: ["AI安全","LoDNA","低资源语言","多语言模型","安全对齐","隐藏状态分析"]
---

# 低资源语言中的跨语言安全对齐幻象

## AI 点评

> 我认为这篇论文戳破了一个危险的幻觉：安全对齐并不会因为模型‘懂’英语就自动迁移到非洲语言。对多语言大模型团队和 AI 安全治理者来说，这是必须正视的警告——把低资源语言纳入安全测试不是锦上添花，而是责任。

## 速览

安全对齐主要基于英语，默认跨语言迁移。该研究用新数据集 LoDNA（直译+本地化提示）和几何框架探测四种非洲语言模型隐藏状态。结果发现跨语言安全迁移受限：有害提示保留不足10%的英文拒绝信号；直译与本地化提示语义高度一致但表征跨层漂移，说明模型编码概念但未触发安全机制，多语言安全对齐表面化。

## 关键点

- 跨语言安全迁移严重不足，有害提示在多数语言-模型组合中保留的英文拒绝信号不足10%。
- 提出 LoDNA 数据集，包含直译和文化本地化提示对，覆盖 Twi、Hausa、Amharic、Swahili 四种非洲语言。
- 提出潜在几何框架，通过隐藏状态中的拒绝表示进行安全评估，超越文本生成式评测。
- 直译与本地化提示语义高度对齐但表征跨层漂移，说明模型未将概念路由到安全机制。

## 全文

Safety alignment in large language models (LLMs) is largely developed in English, assuming these safeguards generalize across multilingual settings. However, this assumption remains underexplored and exposes a vulnerability in low-resource languages. We investigate cross-lingual safety transfer in four African languages, Twi, Hausa, Amharic, and Swahili, using LoDNA, a new safety dataset that pairs literal translations with culturally localized prompts. To move beyond generation-based evaluation, we propose a latent geometric framework that probes hidden-state refusal representations in LLMs. Our experimental results show that cross-lingual safety transfer is severely limited; harmful prompts retain less than 10% of the English refusal signal across most language-model pairs. Literal and localized prompts are semantically aligned (cosine 0.95-0.996) but drift across layers, suggesting models encode the concepts without routing them to safety mechanisms. These findings demonstrate that current multilingual safety alignment is superficial, providing strong evidence against the assumption of a universal, language-agnostic harm manifold within the specific low-resource languages studied. Warning: This paper contains example data that may be offensive or harmful.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_8ecaca672ff564d9de75e4c8