mlsanityData to 7 Oct 2026

Safe at the Margins: A General Approach to Safety Alignment in Low-Resource English Languages -- A Singlish Case Study

arXiv 2502.12485 ↗First posted 18 Feb 2025cs.CL cs.AIcited by 6

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

also in 1 more
Abstract (arXiv)

Ensuring the safety of Large Language Models (LLMs) in diverse linguistic settings remains challenging, particularly for low-resource languages. Existing safety alignment methods are English-centric, limiting their effectiveness. We systematically compare Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Kahneman-Tversky Optimization (KTO) for aligning SEA-Lion-v2.1-Instruct, a Llama 3-8B variant, to reduce toxicity in Singlish. Our results show that SFT+KTO achieves superior safety alignment with higher sample efficiency than DPO. Additionally, we introduce KTO-S, which enhances stability via improved KL divergence regularization. Our approach reduces Singlish toxicity by 99%, generalizes to TOXIGEN, and maintains strong performance on standard LLM benchmarks, providing a scalable framework for safer AI deployment in multilingual contexts.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Placed in Safety alignment and jailbreak attacks, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 17 of its 27 references that are in the map. First posted on arXiv 18 Feb 2025.

Builds on

17 of 27 references.

Closest ancestors · 2

Same direction, newest first

  1. 2024How Alignment and Jailbreak Work: Explain LLM Safety through Intermediate Hidden States9 Jun
  2. 2023Universal and Transferable Adversarial Attacks on Aligned Language Models27 Jul

Other references · 15

  1. 2022Training language models to follow instructions with human feedbackcited by 15,712
  2. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  4. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  5. 2019Fine-Tuning Language Models from Human Preferencescited by 2,124
  6. 2023Jailbroken: How Does LLM Safety Training Fail?cited by 1,565
  7. 2022Red Teaming Language Models with Language Modelscited by 983
  8. 2024Refusal in Language Models Is Mediated by a Single Directioncited by 960
  9. 2023A General Theoretical Paradigm to Understand Learning from Human Preferencescited by 906
  10. 2023Towards Measuring the Representation of Subjective Global Opinions in Language Modelscited by 424
  11. 2024Contrastive Preference Optimization: Pushing the Boundaries of LLM Performance in Machine Translationcited by 390
  12. 2023Low-Resource Languages Jailbreak GPT-4cited by 290
3 more
  1. 2024Is DPO Superior to PPO for LLM Alignment? A Comprehensive Studycited by 262
  2. 2024Unintended Impacts of LLM Alignment on Global Representationcited by 66
  3. 2024LionGuard: Building a Contextualized Moderation Classifier to Tackle Localized Unsafe Contentcited by 5

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 4; 2026: 2.

6 papers.

  1. 2025A Technical Survey of Reinforcement Learning Techniques for Large Language Modelscited by 19
  2. 2025Efficient LLM Moderation with Multi-Layer Latent Prototypescited by 4
  3. 2025Beyond Weaponization: NLP Security for Medium and Lower-Resourced Languages in Their Own Rightcited by 2
  4. 2025Lost in Localization: Building RabakBench with Human-in-the-Loop Validation to Measure Multilingual Safety Gapscited by 2
  5. 2026GR-SAP: Generative Replay for Safety Alignment Preservation during Fine-Tuningcited by 2
  6. 2026LLM Safety Alignment in Low-Resource Languages: A Systematic Literature Reviewcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Towards Comprehensive Post Safety Alignment of Large Language Models via Safety Patching
  2. 2026Multilingual Safety Alignment Via Sparse Weight Editing
  3. 2026Safety at One Shot: Patching Fine-Tuned LLMs with A Single Instance
  4. 2026Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order Optimization
  5. 2025MPO: Multilingual Safety Alignment via Reward Gap Optimization
  6. 2026Can Safety Emerge from Weak Supervision? A Systematic Analysis of Small Language Models
  7. 2025SaLoRA: Safety-Alignment Preserved Low-Rank Adaptation
  8. 2025SafeMERGE: Preserving Safety Alignment in Fine-Tuned Large Language Models via Selective Layer-Wise Model Merging