mlsanityData to 14 Oct 2026

PolyGuard: A Multilingual Safety Moderation Tool for 17 Languages

arXiv 2504.04377 ↗First posted 6 Apr 2025cs.CLcited by 59

Not on the map. Why?

Cited by 40 in LLM safety and security, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

Truly multilingual safety moderation efforts for Large Language Models (LLMs) have been hindered by a narrow focus on a small set of languages (e.g., English, Chinese) as well as a limited scope of safety definition, resulting in significant gaps in moderation capabilities. To bridge these gaps, we release POLYGUARD, a new state-of-the-art multilingual safety model for safeguarding LLM generations, and the corresponding training and evaluation datasets. POLYGUARD is trained on POLYGUARDMIX, the largest multilingual safety training corpus to date containing 1.91M samples across 17 languages (e.g., Chinese, Czech, English, Hindi). We also introduce POLYGUARDPROMPTS, a high quality multilingual benchmark with 29K samples for the evaluation of safety guardrails. Created by combining naturally occurring multilingual human-LLM interactions and human-verified machine translations of an English-only safety dataset (WildGuardMix; Han et al., 2024), our datasets contain prompt-output pairs with labels of prompt harmfulness, response harmfulness, and response refusal. Through extensive evaluations across multiple safety and toxicity benchmarks, we demonstrate that POLYGUARD outperforms existing state-of-the-art open-weight and commercial safety classifiers by 5.5%. Our contributions advance efforts toward safer multilingual LLMs for all global users.

Key facts. Cited by 59 papers in the map (data through 14 Oct 2026). Builds on 45 of its 57 references that are in the map. First posted on arXiv 6 Apr 2025.

Builds on

45 of 57 references.

References · 45

  1. 2024The Llama 3 Herd of Modelscited by 14,715
  2. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  3. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  4. 2024Qwen2.5 Technical Reportcited by 4,942
  5. 2019Unsupervised Cross-lingual Representation Learning at Scalecited by 3,935
  6. 2024HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusalcited by 1,429
  7. 2023Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversationscited by 1,207
  8. 2023BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Datasetcited by 860
  9. 2024WildChat: 1M ChatGPT Interaction Logs in the Wildcited by 629
  10. 2024WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMscited by 496
  11. 2021Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpuscited by 492
  12. 2023LMSYS-Chat-1M: A Large-Scale Real-World LLM Conversation Datasetcited by 461
33 more
  1. 2019Social Bias Frames: Reasoning about Social and Power Implications of Languagecited by 380
  2. 2022A Holistic Approach to Undesired Content Detection in the Real Worldcited by 356
  3. 2019Massively Multilingual Neural Machine Translationcited by 246
  4. 2024ShieldGemma: Generative AI Content Moderation Based on Gemmacited by 243
  5. 2023Multilingual Jailbreak Challenges in Large Language Modelscited by 240
  6. 2024SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusalcited by 240
  7. 2023ToxicChat: Unveiling Hidden Challenges of Toxicity Detection in Real-World User-AI Conversationcited by 231
  8. 2024SALAD-Bench: A Hierarchical and Comprehensive Safety Benchmark for Large Language Modelscited by 228
  9. 2024Tower: An Open Multilingual Large Language Model for Translation-Related Taskscited by 206
  10. 2023Understanding Catastrophic Forgetting in Language Models via Implicit Inferencecited by 144
  11. 2024AEGIS: Online Adaptive AI Content Safety Moderation with Ensemble of LLM Expertscited by 136
  12. 2025Aegis2.0: A Diverse AI Safety Dataset and Risks Taxonomy for Alignment of LLM Guardrailscited by 132
  13. 2024A Survey on Multilingual Large Language Models: Corpora, Alignment, and Biascited by 84
  14. 2024AIR-Bench 2024: A Safety Benchmark Based on Risk Categories from Regulations and Policiescited by 74
  15. 2024The Multilingual Alignment Prism: Aligning Global and Local Preferences to Reduce Harmcited by 70
  16. 2025International AI Safety Reportcited by 67
  17. 2024Refusal-Trained LLMs Are Easily Jailbroken As Browser Agentscited by 65
  18. 2023All Languages Matter: On the Multilingual Safety of Large Language Modelscited by 53
  19. 2024RTP-LX: Can LLMs Evaluate Toxicity in Multilingual Scenarios?cited by 36
  20. 2023COBRA Frames: Contextual Reasoning about Effects and Harms of Offensive Statementscited by 34
  21. 2024Code-Switching Red-Teaming: LLM Evaluation for Safety and Multilingual Understandingcited by 34
  22. 2024AI Safety in Generative AI Large Language Models: A Surveycited by 33
  23. 2022Toxicity in Multilingual Machine Translation at Scalecited by 26
  24. 2025Lessons From Red Teaming 100 Generative AI Productscited by 23
  25. 2024PolygloToxicityPrompts: Multilingual Evaluation of Neural Toxic Degeneration in Large Language Modelscited by 20
  26. 2025Enhancing LLM Safety Through a Theoretical Minimax Game Lenscited by 19
  27. 2024CHiSafetyBench: A Chinese Hierarchical Safety Benchmark for Large Language Modelscited by 15
  28. 2024Benchmarking LLM Guardrails in Handling Multilingual Toxicitycited by 14
  29. 2024Multilingual Large Language Models and Curse of Multilingualitycited by 13
  30. 2024AI Safety Landscape for Large Language Models: Taxonomy, State-of-the-art, and Future Directionscited by 8
  31. 2024ChineseSafe: A Chinese Benchmark for Evaluating Safety in Large Language Modelscited by 7
  32. 2024Chinese SafetyQA: A Safety Short-form Factuality Benchmark for Large Language Modelscited by 7
  33. 2024GeniL: A Multilingual Dataset on Generalizing Languagecited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 10; 2026: 49.

59 papers.

Sort:

By area:

  1. 2025Qwen3Guard Technical Reportcited by 155
  2. 2026ToolSafe: Enhancing Tool Invocation Safety of LLM-based agents via Proactive Step-level Guardrail and Feedbackcited by 72
  3. 2025CultureGuard: Towards Culturally-Aware Dataset and Guard Model for Multilingual Safety Applicationscited by 27
  4. 2026YuFeng-XGuard: A Reasoning-Centric, Interpretable, and Flexible Guardrail Model for Large Language Modelscited by 19
  5. 2026LLM Safety From Within: Detecting Harmful Content with Internal Representationscited by 11
  6. 2025mR3: Multilingual Rubric-Agnostic Reward Reasoning Modelscited by 8
  7. 2025OMNIGUARD: An Efficient Approach for AI Safety Moderation Across Languages and Modalitiescited by 7
  8. 2026StepShield: When, Not Whether to Intervene on Rogue Agentscited by 7
  9. 2026Selective Steering: Norm-Preserving Control Through Discriminative Layer Selectioncited by 6
  10. 2026SingGuard: A Policy-Adaptive Multimodal LLM Guardrail with Dynamic Reasoningcited by 6
  11. 2025FanarGuard: A Culturally-Aware Moderation Filter for Arabic Language Modelscited by 5
  12. 2026UbuntuGuard: A Culturally-Grounded Policy Benchmark for Equitable AI Safety in African Languagescited by 5
38 more
  1. 2026LPG: Balancing Efficiency and Policy Reasoning in Latent Policy Guardrailscited by 5
  2. 2026Unveiling Language Routing Isolation in Multilingual MoE Models for Interpretable Subnetwork Adaptationcited by 4
  3. 2026GLiGuard: Schema-Conditioned Classification for LLM Safeguardcited by 4
  4. 2026MindGuard: Guardrail Classifiers for Multi-Turn Mental Health Supportcited by 3
  5. 2026CAGE: A Framework for Culturally Adaptive Red-Teaming Benchmark Generationcited by 3
  6. 2026Predict, Don't React: Value-Based Safety Forecasting for LLM Streamingcited by 3
  7. 2026LiSA: Lifelong Safety Adaptation via Conservative Policy Inductioncited by 3
  8. 2026TukaBench: A Culturally Grounded Jailbreak Benchmark for African Languagescited by 3
  9. 2026SeerGuard: A Safety Framework for Mobile GUI Agents via World Model Predictioncited by 3
  10. 2025Lost in Localization: Building RabakBench with Human-in-the-Loop Validation to Measure Multilingual Safety Gapscited by 2
  11. 2025CREST: Universal Safety Guardrails Through Cluster-Guided Cross-Lingual Transfercited by 2
  12. 2026SEA-Guard: Culturally Grounded Multilingual Safeguard for Southeast Asiacited by 2
  13. 2026GuardReasoner-Omni: A Reasoning-based Multi-modal Guardrail for Text, Image, Video, and Audiocited by 2
  14. 2026GLiNER Guard: Unified Encoder Family for Production LLM Safety and Privacycited by 2
  15. 2026HARP: Measuring Harm Amplification in Multi-Agent LLM Systemscited by 2
  16. 2026FinGuard: Detecting Financial Regulatory Non-Compliance in LLM Interactionscited by 2
  17. 2026Low-Resource Safety Failures Are Action Failures, Not Representation Failurescited by 2
  18. 2025The Problem with Safety Classification is not just the Modelscited by 1
  19. 2026TWGuard: A Case Study of LLM Safety Guardrails for Localized Linguistic Contextscited by 1
  20. 2026BabelSafe: A Policy-Grounded Multilingual Safety Benchmark for LLMscited by 1
  21. 2026BELLS-O: Evaluating the Operational Trade-offs of LLM Supervision Systemscited by 1
  22. 2026HaloGuard 1.0: An Open Weights Constitutional Classifier for Multilingual AI Safetycited by 1
  23. 2026Shieldstralcited by 1
  24. 2026The Illusion of Cross-Lingual Safety in Low-Resource Languagescited by 1
  25. 2025Adversarial Distilled Retrieval-Augmented Guarding Model for Online Malicious Intent Detectioncited by 0
  26. 2025Guard Vector: Beyond English LLM Guardrails with Task-Vector Composition and Streaming-Aware Prefix SFTcited by 0
  27. 2026Where Does Toxicity Live? Mechanistic Localization and Targeted Suppression in Language Modelscited by 0
  28. 2026Opir: Efficient Multi-Task Safety Classification for Toxicity, Jailbreaks, Hate Speech, and Harmful Contentcited by 0
  29. 2026All you need to break LLMs are Black-Box, Adapting, Efficient, Transferable, Harmful, Applicable ... Attackscited by 0
  30. 2026SpeechJBB: Probing Safety Alignment and Comprehension in Large Audio Language Models under Code-Switched Speechcited by 0
  31. 2026CHILLGuard: Towards Fine-Grained Chinese LLM Safety Guardrail with Scalable Data Construction and Model-aware Preference Alignmentcited by 0
  32. 2026A Survey of Toxicity Detection and Mitigation Strategies for Multilingual Language Modelscited by 0
  33. 2026DT-Guard: Intent-Driven Reasoning-Active Training for Reasoning-Free LLM Safety Guardrailcited by 0
  34. 2026RAIL Guard: Closing the Evaluation-to-Remediation Gap in Responsible AI for LLM Agentscited by 0
  35. 2026ArabicDialectSafety: A Dialect-Aware Benchmark for Arabic Content Safety Classificationcited by 0
  36. 2026Pragmatic Attack Surface: Vulnerabilities of Implicit Context in Large Language Modelscited by 0
  37. 2026ProbGuard: Calibrated Safety Risk Estimation from LLM Output Distributionscited by 0
  38. 2026Policy-as-logic for robust reasoning over rulescited by 0

50 most cited of 59 All 59 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025X-Guard: Multilingual Guard Agent for Content Moderation
  2. 2026IndicGuard: A Multilingual Safety Guard Model and Dataset for Indic Languages
  3. 2025The State of Multilingual LLM Safety Research: From Measuring the Language Gap to Mitigating It
  4. 2025LinguaSafe: A Comprehensive Multilingual Safety Benchmark for Large Language Models
  5. 2026IndicSafe: A Benchmark for Evaluating Multilingual LLM Safety in South Asia
  6. 2025MrGuard: A Multilingual Reasoning Guardrail for Universal LLM Safety
  7. 2026Sch"utzen: Evaluating LLM Safety in Bulgarian and German Contexts
  8. 2025Toxicity Red-Teaming: Benchmarking LLM Safety in Singapore's Low-Resource Languages