LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety

• URL: http://arxiv.org/abs/2604.12710v1
• Date: Mon, 13 Apr 2026 15:59:50 GMT
• Title: LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety
• Abstract summary: Large language models (LLMs) often demonstrate strong safety performance in high-resource languages, yet exhibit severe vulnerabilities in low-resource languages.<n>We attribute this gap to a mismatch between language-agnostic semantic understanding ability and language-dominant safety alignment biased toward high-resource languages.<n>We propose Language-Agnostic Semantic Alignment (LASA), which anchors safety alignment directly in semantic bottlenecks.
• Score: 88.98698230989186
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large language models (LLMs) often demonstrate strong safety performance in high-resource languages, yet exhibit severe vulnerabilities when queried in low-resource languages. We attribute this gap to a mismatch between language-agnostic semantic understanding ability and language-dominant safety alignment biased toward high-resource languages. Consistent with this hypothesis, we empirically identify the semantic bottleneck in LLMs, an intermediate layer in which the geometry of model representations is governed primarily by shared semantic content rather than language identity. Building on this observation, we propose Language-Agnostic Semantic Alignment (LASA), which anchors safety alignment directly in semantic bottlenecks. Experiments show that LASA substantially improves safety across all languages: average attack success rate (ASR) drops from 24.7% to 2.8% on LLaMA-3.1-8B-Instruct and remains around 3-4% across Qwen2.5 and Qwen3 Instruct models (7B-32B). Together, our analysis and method offer a representation-level perspective on LLM safety, suggesting that safety alignment requires anchoring safety understanding not in surface text, but in the model's language-agnostic semantic space.

Related papers

• When Safety Speaks a Language: A Mechanistic Analysis of Safety-Language Identity Entanglement in LLMs [4.92475688183711]
  Safety-relevant features are architecture-dependent in terms of where they are located and how they are distributed across layers.<n>This safety-language entanglement has direct consequences such that ablating safety features impacts not only harmful response rates but also target language.<n>Our findings qualify the language-universality of safety alignment as architecture-dependent and offer a mechanistic account of multilingual safety interventions.
  arXiv  Detail & Related papers  (2026-08-30T18:07:26Z)
• LLM Safety Alignment in Low-Resource Languages: A Systematic Literature Review [9.895109697164283]
  Large Language Models (LLMs) have achieved substantial progress in safety alignment.<n>Their safety guarantees remain significantly weaker in low-resource and multilingual settings.
  arXiv  Detail & Related papers  (2026-07-20T21:11:00Z)
• Multilingual Refusal Alignment for Safer Large Language Models [53.64286756804503]
  We investigate whether single-language alignment transfers cross-lingually, how language consistency is preserved during training, and the resulting trade-offs with general knowledge capabilities.<n>We introduce RefusEU, a novel refusal alignment dataset covering 12 European languages, including a dedicated test set for evaluating current state-of-the-art models.<n>Our controlled Direct Preference Optimization (DPO) experiments provide two key insights: aligning models exclusively in English is insufficient to ensure cross-lingual safety, even for the same harm categories, whereas training on multilingual datasets can improve safety without degrading general performance, as measured by the Global MMLU benchmark.
  arXiv  Detail & Related papers  (2026-04-24T09:29:14Z)
• Precise Shield: Explaining and Aligning VLLM Safety via Neuron-Level Guidance [64.18179414483966]
  In real-world deployments, Vision-Language Large Models (VLLMs) face critical challenges from multilingual and multimodal composite attacks.<n>We propose Precise Shield, a two-stage framework that first identifies safety neurons by contrasting activation patterns between harmful and benign inputs.<n>This strategy substantially improves safety while preserving multilingual and multimodal generalization.
  arXiv  Detail & Related papers  (2026-04-10T02:42:52Z)
• IndicSafe: A Benchmark for Evaluating Multilingual LLM Safety in South Asia [0.6978180153516672]
  We present the first systematic evaluation of large language models (LLMs) safety across 12 Indic languages.<n>Cross-language agreement is just 12.8%, and textttSAFE rate variance exceeds 17% across languages.<n>We release textscIndicSafe, the first benchmark to enable culturally informed safety evaluation for Indic deployments.
  arXiv  Detail & Related papers  (2026-03-18T16:54:07Z)
• SEA-SafeguardBench: Evaluating AI Safety in SEA Languages and Cultures [36.95168918567729]
  Existing multilingual safety benchmarks often rely on machine-translated English data.<n>We introduce SEA-SafeguardBench, the first human-verified safety benchmark for SEA.<n>It covers eight languages, 21,640 samples, across three subsets: general, in-the-wild, and content generation.
  arXiv  Detail & Related papers  (2025-12-05T07:57:57Z)
• Self-Aware Safety Augmentation: Leveraging Internal Semantic Understanding to Enhance Safety in Vision-Language Models [21.961325147038867]
  Large vision-language models (LVLMs) are vulnerable to harmful input compared to their language-only backbones.<n>We define these capabilities as safety perception, semantic understanding, and alignment for linguistic expression.<n>Motivated by these findings, we propose textbfSelf-Aware Safety Augmentation (SASA), a technique that projects informative semantic representations onto earlier safety-oriented layers.
  arXiv  Detail & Related papers  (2025-07-29T09:48:57Z)
• MPO: Multilingual Safety Alignment via Reward Gap Optimization [88.76638442683391]
  Large language models (LLMs) have become increasingly central to AI applications worldwide.<n>Existing preference learning methods for safety alignment, such as RLHF and DPO, are primarily monolingual and struggle with noisy multilingual data.<n>We introduce Multilingual reward gaP Optimization (MPO), a novel approach that leverages the well-aligned safety capabilities of the dominant language (English) to improve safety alignment across multiple languages.
  arXiv  Detail & Related papers  (2025-05-22T16:24:51Z)
• The Hidden Space of Safety: Understanding Preference-Tuned LLMs in Multilingual context [0.9130277390156759]
  Alignment tuning has enabled large language models to excel in reasoning, instruction-following, and minimizing harmful generations.<n>Despite their widespread deployment, these models exhibit a monolingual bias, raising concerns about the effectiveness of alignment across languages.<n>Current alignment methods predominantly focus on English, leaving it unclear how alignment mechanism generalizes to multilingual settings.
  arXiv  Detail & Related papers  (2025-04-03T15:46:46Z)
• LLMs Lost in Translation: M-ALERT uncovers Cross-Linguistic Safety Inconsistencies [63.10843814055688]
  M-ALERT is a benchmark that evaluates the safety of Large Language Models in five languages.<n>M-ALERT includes 15k high-quality prompts per language, totaling 75k, with category-wise annotations.<n>Our experiments on 39 state-of-the-art LLMs highlight the importance of language-specific safety analysis.
  arXiv  Detail & Related papers  (2024-12-19T16:46:54Z)
• All Languages Matter: On the Multilingual Safety of Large Language Models [96.47607891042523]
  We build the first multilingual safety benchmark for large language models (LLMs) XSafety covers 14 kinds of commonly used safety issues across 10 languages that span several language families. We propose several simple and effective prompting methods to improve the multilingual safety of ChatGPT.
  arXiv  Detail & Related papers  (2023-10-02T05:23:34Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.