mlsanityData to 14 Oct 2026

Multilingual Safety Alignment Via Sparse Weight Editing

arXiv 2602.22554 ↗First posted 26 Feb 2026cs.LGcited by 6

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

also in 1 more
Abstract (arXiv)

Large Language Models (LLMs) exhibit significant safety disparities across languages, with low-resource languages (LRLs) often bypassing safety guardrails established for high-resource languages (HRLs) like English. Existing solutions, such as multilingual supervised fine-tuning (SFT) or Reinforcement Learning from Human Feedback (RLHF), are computationally expensive and dependent on scarce multilingual safety data. In this work, we propose a novel, training-free alignment framework based on Sparse Weight Editing. Identifying that safety capabilities are localized within a sparse set of safety neurons, we formulate the cross-lingual alignment problem as a constrained linear transformation. We derive a closed-form solution to optimally map the harmful representations of LRLs to the robust safety subspaces of HRLs, while preserving general utility via a null-space projection constraint. Extensive experiments across 8 languages and multiple model families (Llama-3, Qwen-2.5) demonstrate that our method substantially reduces Attack Success Rate (ASR) in LRLs with negligible impact on general reasoning capabilities, all achieved with a single, data-efficient calculation.

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Placed in Safety alignment and jailbreak attacks, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 36 of its 41 references that are in the map. First posted on arXiv 26 Feb 2026.

Builds on

36 of 41 references.

Closest ancestors · 5

Same direction, newest first

  1. 2026SafeNeuron: Neuron-Level Safety Alignment for Large Language Models12 Feb
  2. 2025NeuroStrike: Neuron-Level Attacks on Aligned LLMs15 Sep
  3. 2025MPO: Multilingual Safety Alignment via Reward Gap Optimization22 May
  4. 2024Language Models are Homer Simpson! Safety Re-Alignment of Fine-tuned Language Models through Task Arithmetic19 Feb
  5. 2023Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrations10 Oct

Other references · 31

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2022Training language models to follow instructions with human feedbackcited by 15,971
  3. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  4. 2025Qwen3 Technical Reportcited by 8,820
  5. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  6. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  7. 2013Intriguing properties of neural networkscited by 5,155
  8. 2021Finetuned Language Models Are Zero-Shot Learnerscited by 3,780
  9. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,720
  10. 2023Jailbreaking Black Box Large Language Models in Twenty Queriescited by 1,420
  11. 2023Tree of Attacks: Jailbreaking Black-Box LLMs Automaticallycited by 672
  12. 2021Knowledge Neurons in Pretrained Transformerscited by 601
19 more
  1. 2018Scalable agent alignment via reward modeling: a research directioncited by 447
  2. 2024Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Modelscited by 435
  3. 2024A StrongREJECT for Empty Jailbreakscited by 385
  4. 2023DeepInception: Hypnotize Large Language Model to Be Jailbreakercited by 367
  5. 2023AI Alignment: A Comprehensive Surveycited by 263
  6. 2024Jailbreak Attacks and Defenses Against Large Language Models: A Surveycited by 230
  7. 2024Transcoders Find Interpretable LLM Feature Circuitscited by 200
  8. 2024AutoDAN-Turbo: A Lifelong Agent for Strategy Self-Exploration to Jailbreak LLMscited by 199
  9. 2021Alignment of Language Agentscited by 168
  10. 2025Qwen3Guard Technical Reportcited by 155
  11. 2023Do Multilingual Language Models Think Better in English?cited by 88
  12. 2022Finding Skill Neurons in Pre-trained Transformer-based Language Modelscited by 66
  13. 2023Revisiting Machine Translation for Cross-lingual Classificationcited by 41
  14. 2024AdaMergeX: Cross-Lingual Transfer with Large Language Models via Adaptive Adapter Mergingcited by 34
  15. 2021Modelling Latent Translations for Cross-Lingual Transfercited by 16
  16. 2024Preventing Catastrophic Overfitting in Fast Adversarial Training: A Bi-level Optimization Perspectivecited by 13
  17. 2025Implicit Jailbreak Attacks via Cross-Modal Information Concealment on Vision-Language Modelscited by 6
  18. 2025AlignX: Advancing Multilingual Large Language Models with Multilingual Representation Alignmentcited by 6
  19. 2025Linguistic Neuron Overlap Patterns to Facilitate Cross-lingual Transfer on Low-resource Languagescited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 6.

6 papers.

  1. 2026Multilingual Safety Alignment via Self-Distillationcited by 3
  2. 2026Low-Resource Safety Failures Are Action Failures, Not Representation Failurescited by 2
  3. 2026Targeted Interpretable Safety Neuron Enhancement for Multilingual Vision-Language Large Modelscited by 0
  4. 2026One Anchor for All: Unified Multilingual and Multimodal Safety Alignment for LVLMscited by 0
  5. 2026LLM Safety Alignment in Low-Resource Languages: A Systematic Literature Reviewcited by 0
  6. 2026When Safety Speaks a Language: A Mechanistic Analysis of Safety-Language Identity Entanglement in LLMscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Layer-wise Swapping for Generalizable Multilingual Safety
  2. 2026BabelSteering: Multilingual Safety Alignment via English Steering Vectors
  3. 2026Preventing Safety Drift in Large Language Models via Coupled Weight and Activation Constraints
  4. 2025Safe at the Margins: A General Approach to Safety Alignment in Low-Resource English Languages -- A Singlish Case Study
  5. 2024The Language Barrier: Dissecting Safety Challenges of LLMs in Multilingual Contexts
  6. 2026Safety at One Shot: Patching Fine-Tuned LLMs with A Single Instance
  7. 2024One-Shot Safety Alignment for Large Language Models via Optimal Dualization
  8. 2026SafeSteer: Localized On-Policy Distillation for Efficient Safety Alignment