mlsanityData to 13 Oct 2026

Align Once, Benefit Multilingually: Enforcing Multilingual Consistency for LLM Safety Alignment

arXiv 2602.16660 ↗First posted 18 Feb 2026cs.CL cs.AI cs.LGcited by 9

On the map: LLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks

also in 1 more
Abstract (arXiv)

The widespread deployment of large language models (LLMs) across linguistic communities necessitates reliable multilingual safety alignment. However, recent efforts to extend alignment to other languages often require substantial resources, either through large-scale, high-quality supervision in the target language or through pairwise alignment with high-resource languages, which limits scalability. In this work, we propose a resource-efficient method for improving multilingual safety alignment. We introduce a plug-and-play Multi-Lingual Consistency (MLC) loss that can be integrated into existing monolingual alignment pipelines. By improving collinearity between multilingual representation vectors, our method encourages directional consistency at the multilingual semantic level in a single update. This allows simultaneous alignment across multiple languages using only multilingual prompt variants without requiring additional response-level supervision in low-resource languages. We validate the proposed method across different model architectures and alignment paradigms, and demonstrate its effectiveness in enhancing multilingual safety with limited impact on general model utility. Further evaluation across languages and tasks indicates improved cross-lingual generalization, suggesting the proposed approach as a practical solution for multilingual consistency alignment under limited supervision.

Key facts. Cited by 9 papers in the map (data through 13 Oct 2026). Placed in Safety alignment and jailbreak attacks, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 43 of its 52 references that are in the map. First posted on arXiv 18 Feb 2026.

Builds on

43 of 52 references.

Closest ancestors · 5

Same direction, newest first

  1. 2025MPO: Multilingual Safety Alignment via Reward Gap Optimization22 May
  2. 2024PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference20 Jun
  3. 2024Safety Alignment Should Be Made More Than Just a Few Tokens Deep10 Jun
  4. 2024The Language Barrier: Dissecting Safety Challenges of LLMs in Multilingual Contexts23 Jan
  5. 2023Safe RLHF: Safe Reinforcement Learning from Human Feedback19 Oct

Other references · 38

  1. 2022Training language models to follow instructions with human feedbackcited by 15,856
  2. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,979
  3. 2025Qwen3 Technical Reportcited by 8,510
  4. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,986
  5. 2020Measuring Massive Multitask Language Understandingcited by 7,876
  6. 2022BLOOM: A 176B-Parameter Open-Access Multilingual Language Modelcited by 1,978
  7. 2024LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Modelscited by 1,450
  8. 2022No Language Left Behind: Scaling Human-Centered Machine Translationcited by 1,302
  9. 2019Cross-lingual Language Model Pretrainingcited by 1,089
  10. 2024SimPO: Simple Preference Optimization with a Reference-Free Rewardcited by 954
  11. 2021The FLORES-101 Evaluation Benchmark for Low-Resource and Multilingual Machine Translationcited by 556
  12. 2020MAD-X: An Adapter-Based Framework for Multi-Task Cross-Lingual Transfercited by 443
26 more
  1. 2023Efficient and Effective Text Encoding for Chinese LLaMA and Alpacacited by 304
  2. 2023Multilingual Jailbreak Challenges in Large Language Modelscited by 235
  3. 2024How do Large Language Models Handle Multilingualism?cited by 150
  4. 2024Language-Specific Neurons: The Key to Multilingual Capabilities in Large Language Modelscited by 125
  5. 2023Don't Trust ChatGPT when Your Question is not in English: A Study of Multilingual Abilities and Types of LLMscited by 104
  6. 2017Offline bilingual word vectors, orthogonal transformations and the inverted softmaxcited by 95
  7. 2024LLaMA Beyond English: An Empirical Study on Language Capability Transfercited by 94
  8. 2024MAPO: Advancing Multilingual Reasoning through Multilingual Alignment-as-Preference Optimizationcited by 69
  9. 2024The Semantic Hub Hypothesis: Language Models Share Semantic Representations Across Languages and Modalitiescited by 69
  10. 2024Clio: Privacy-Preserving Insights into Real-World AI Usecited by 69
  11. 2023All Languages Matter: On the Multilingual Safety of Large Language Modelscited by 53
  12. 2025The State of Multilingual LLM Safety Research: From Measuring the Language Gap to Mitigating Itcited by 43
  13. 2024Enhancing Multilingual Capabilities of Large Language Models through Self-Distillation from Resource-Rich Languagescited by 36
  14. 2023Improving In-context Learning of Multilingual Generative Language Models with Cross-lingual Alignmentcited by 33
  15. 2024Language Imbalance Driven Rewarding for Multilingual Self-improvingcited by 27
  16. 2023LLaMAntino: LLaMA 2 Models for Effective Text Generation in Italian Languagecited by 24
  17. 2023InstructAlign: High-and-Low Resource Language Alignment via Continual Crosslingual Instruction Tuningcited by 21
  18. 2024Beneath the Surface of Consistency: Exploring Cross-lingual Knowledge Representation Sharing in LLMscited by 20
  19. 2025Language-Specific Latent Process Hinders Cross-Lingual Performancecited by 20
  20. 2023A Material Lens on Coloniality in NLPcited by 19
  21. 2025Continual Multimodal Contrastive Learningcited by 19
  22. 2025Principled Multimodal Representation Learningcited by 19
  23. 2024Discovering Low-rank Subspaces for Language-agnostic Multilingual Representationscited by 18
  24. 2024MLaKE: Multilingual Knowledge Editing Benchmark for Large Language Modelscited by 11
  25. 2024BayLing 2: A Multilingual Large Language Model with Efficient Language Alignmentcited by 8
  26. 2024Lens: Rethinking Multilingual Enhancement for Large Language Modelscited by 5

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2026: 9.

9 papers.

  1. 2026Multilingual Safety Alignment via Self-Distillationcited by 3
  2. 2026Low-Resource Safety Failures Are Action Failures, Not Representation Failurescited by 2
  3. 2026Sycophancy as a Multilingual Alignment Failure: How Safety Degrades Across Languages, Topics, and Modelscited by 0
  4. 2026Adversarial Concept Search: Predicting Compositional Errors From Feature Geometrycited by 0
  5. 2026One Anchor for All: Unified Multilingual and Multimodal Safety Alignment for LVLMscited by 0
  6. 2026LLM Safety Alignment in Low-Resource Languages: A Systematic Literature Reviewcited by 0
  7. 2026Latent Space Refusal Anchoring for Low-Resource African Languages: Mechanistic Safety Recovery Without Retrainingcited by 0
  8. 2026When Safety Speaks a Language: A Mechanistic Analysis of Safety-Language Identity Entanglement in LLMscited by 0
  9. 2026Multilingual Safety Signals Are Multi-Layered: Filtering Safety-Degrading Data for Safer LLMscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025The Hidden Space of Safety: Understanding Preference-Tuned LLMs in Multilingual context
  2. 2024Multilingual Blending: LLM Safety Alignment Evaluation with Language Mixture
  3. 2025AlignX: Advancing Multilingual Large Language Models with Multilingual Representation Alignment
  4. 2025CM-Align: Consistency-based Multilingual Alignment for Large Language Models
  5. 2026Multilingual Refusal Alignment for Safer Large Language Models
  6. 2025Middle-Layer Representation Alignment for Cross-Lingual Transfer in Fine-Tuned LLMs
  7. 2024Concept Space Alignment in Multilingual LLMs
  8. 2026Multilingual Safety Alignment Via Sparse Weight Editing