# LionGuard 2: Building Lightweight, Data-Efficient & Localised Multilingual Content Moderators

arXiv: https://arxiv.org/abs/2507.15339 · HTML: https://mlsanity.com/paper/2507.15339/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 45 references that are in the map.
- First posted on arXiv 21 Jul 2025.

## Builds on (most cited first, 18 of 18)

- [Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversations](https://mlsanity.com/paper/2312.06674/) (2023)
- [M3-Embedding: Multi-Linguality, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillation](https://mlsanity.com/paper/2402.03216/) (2024)
- [BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset](https://mlsanity.com/paper/2307.04657/) (2023)
- [WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs](https://mlsanity.com/paper/2406.18495/) (2024)
- [Low-Resource Languages Jailbreak GPT-4](https://mlsanity.com/paper/2310.02446/) (2023)
- [jina-embeddings-v3: Multilingual Embeddings With Task LoRA](https://mlsanity.com/paper/2409.10173/) (2024)
- [Aegis2.0: A Diverse AI Safety Dataset and Risks Taxonomy for Alignment of LLM Guardrails](https://mlsanity.com/paper/2501.09004/) (2025)
- [Llama Guard 3 Vision: Safeguarding Human-AI Image Understanding Conversations](https://mlsanity.com/paper/2411.10414/) (2024)
- [The Language Barrier: Dissecting Safety Challenges of LLMs in Multilingual Contexts](https://mlsanity.com/paper/2401.13136/) (2024)
- [The Alternative Annotator Test for LLM-as-a-Judge: How to Statistically Justify Replacing Human Annotators with LLMs](https://mlsanity.com/paper/2501.10970/) (2025)
- [The Amazon Nova Family of Models: Technical Report and Model Card](https://mlsanity.com/paper/2506.12103/) (2025)
- [Recent Advances in Hate Speech Moderation: Multimodality and the Role of Large Models](https://mlsanity.com/paper/2401.16727/) (2024)
- [Llama Guard 3-1B-INT4: Compact and Efficient Safeguard for Human-AI Conversations](https://mlsanity.com/paper/2411.17713/) (2024)
- [Enhancing LLM Safety Through a Theoretical Minimax Game Lens](https://mlsanity.com/paper/2502.05163/) (2025)
- [SGHateCheck: Functional Tests for Detecting Hate Speech in Low-Resource Languages of Singapore](https://mlsanity.com/paper/2405.01842/) (2024)
- [Disentangling Singlish Discourse Particles with Task-Driven Representation](https://mlsanity.com/paper/2409.20366/) (2024)
- [Measuring What Matters: A Framework for Evaluating Safety Risks in Real-World LLM Applications](https://mlsanity.com/paper/2507.09820/) (2025)
- [Social Cyber Geographical Worldwide Inventory of Bots](https://mlsanity.com/paper/2501.18839/) (2025)

## Built on by (most cited first, 7 of 7)

- [Cross-Lingual Sentiment Misalignment: Auditing Multilingual Language Models for Inversion Risk, Dialectal Representation, and Affective Stability](https://mlsanity.com/paper/2602.17469/) (2026)
- [Toxicity Red-Teaming: Benchmarking LLM Safety in Singapore's Low-Resource Languages](https://mlsanity.com/paper/2509.15260/) (2025)
- [SEA-Guard: Culturally Grounded Multilingual Safeguard for Southeast Asia](https://mlsanity.com/paper/2602.01618/) (2026)
- [SEA-SafeguardBench: Evaluating AI Safety in SEA Languages and Cultures](https://mlsanity.com/paper/2512.05501/) (2025)
- [Stylistic Evolution and LLM Neutrality in Singlish Language](https://mlsanity.com/paper/2601.06580/) (2026)
- [Structured Semantic Cloaking for Jailbreak Attacks on Large Language Models](https://mlsanity.com/paper/2603.16192/) (2026)
- [LLM Safety Alignment in Low-Resource Languages: A Systematic Literature Review](https://mlsanity.com/paper/2608.14626/) (2026)

## Abstract (arXiv)

Modern moderation systems increasingly support multiple languages, but often fail to address localisation and low-resource variants - creating safety gaps in real-world deployments. Small models offer a potential alternative to large LLMs, yet still demand considerable data and compute. We present LionGuard 2, a lightweight, multilingual moderation classifier tailored to the Singapore context, supporting English, Chinese, Malay, and partial Tamil. Built on pre-trained OpenAI embeddings and a multi-head ordinal classifier, LionGuard 2 outperforms several commercial and open-source systems across 17 benchmarks, including both Singapore-specific and public English datasets. The system is actively deployed within the Singapore Government, demonstrating practical efficacy at scale. Our findings show that high-quality local data and robust multilingual embeddings can achieve strong moderation performance, without fine-tuning large models. We release our model weights and part of our training data to support future work on LLM safety.