mlsanityData to 14 Oct 2026

Register Shifts Break LLM Safety: A Bengali Benchmark with Culturally Grounded Harms

arXiv 2608.22335 ↗First posted 23 Aug 2026cs.CL cs.AIcited by 1

On the map: LLM safety and security → Safety alignment, refusal and guardrails

Abstract (arXiv)

Bengali is the seventh-most-spoken language globally, yet LLM safety evaluation remains overwhelmingly English-centric. We introduce BanglaSafe, a benchmark of 879 Bengali prompts combining 309 natively authored prompts with 570 expert-reviewed prompts, spanning 17 culturally grounded harm categories and five prompting conditions that vary language, writing style, and authority framing. Evaluating 18 frontier LLMs, we find that over half of all responses are unsafe or partially unsafe (53.6%) while 14.7% contains strictly harmful content, and that the strongest observed effect is not the switch from English to Bengali but the choice of writing style within Bengali: the same harmful request phrased as a formal newspaper investigation succeeds 17 percentage points more often than the same request phrased as a casual message, with no adversarial engineering involved. We further show that existing safety classifiers struggle to reliably evaluate Bengali content, with even frontier models failing on nearly half of all cases.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security. Builds on 31 of its 58 references that are in the map. First posted on arXiv 23 Aug 2026.

Builds on

31 of 58 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026IndicSafe: A Benchmark for Evaluating Multilingual LLM Safety in South Asia18 Mar
  2. 2025Lost in Localization: Building RabakBench with Human-in-the-Loop Validation to Measure Multilingual Safety Gaps8 Jul
  3. 2024PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference20 Jun
  4. 2024ALERT: A Comprehensive Benchmark for Assessing Large Language Models' Safety through Red Teaming6 Apr

Other references · 27

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
  3. 2024HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusalcited by 1,429
  4. 2026DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligencecited by 1,040
  5. 2023BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Datasetcited by 860
  6. 2024How Johnny Can Persuade LLMs to Jailbreak Them: Rethinking Persuasion to Challenge AI Safety by Humanizing LLMscited by 573
  7. 2023Catastrophic Jailbreak of Open-source LLMs via Exploiting Generationcited by 451
  8. 2023Low-Resource Languages Jailbreak GPT-4cited by 309
  9. 2023Scalable and Transferable Black-Box Jailbreaks for Language Models via Persona Modulationcited by 214
  10. 2024Adding Error Bars to Evals: A Statistical Approach to Language Model Evaluationscited by 148
  11. 2024The Language Barrier: Dissecting Safety Challenges of LLMs in Multilingual Contextscited by 94
  12. 2024PARIKSHA: A Large-Scale Investigation of Human-LLM Evaluator Agreement on Multilingual and Multi-Cultural Datacited by 37
15 more
  1. 2024Code-Switching Red-Teaming: LLM Evaluation for Safety and Multilingual Understandingcited by 34
  2. 2025CultureGuard: Towards Culturally-Aware Dataset and Guard Model for Multilingual Safety Applicationscited by 27
  3. 2024Scaling Trends in Language Model Robustnesscited by 20
  4. 2025TigerLLM - A Family of Bangla Large Language Modelscited by 16
  5. 2026DAGGER: Distractor-Aware Graph Generation for Executable Reasoning in Math Problemscited by 9
  6. 2025Read Between the Lines: A Benchmark for Uncovering Political Bias in Bangla News Articlescited by 6
  7. 2026GanitLLM: Difficulty-Aware Bengali Mathematical Reasoning through Curriculum-GRPOcited by 6
  8. 2024BanglaLlama: LLaMA for Bangla Languagecited by 5
  9. 2025If We May De-Presuppose: Robustly Verifying Claims through Presupposition-Free Question Decompositioncited by 5
  10. 2026IndicJR: A Judge-Free Benchmark of Jailbreak Robustness in South Asian Languagescited by 5
  11. 2026Cross-Lingual Sentiment Misalignment: Auditing Multilingual Language Models for Inversion Risk, Dialectal Representation, and Affective Stabilitycited by 5
  12. 2025BanglaTalk: Towards Real-Time Speech Assistance for Bengali Regional Dialectscited by 3
  13. 2026Many Dialects, Many Languages, One Cultural Lens: Evaluating Multilingual VLMs for Bengali Culture Understanding Across Historically Linked Languages and Regional Dialectscited by 3
  14. 2026BanglaIPA: Towards Robust Text-to-IPA Transcription with Contextual Rewriting in Bengalicited by 2
  15. 2025BengaliMoralBench: A Benchmark for Auditing Moral Reasoning in Large Language Models within Bengali Language and Culturecited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026New Evidence, Same Choice: Testing Physical Experiment Selection in Vision Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026SurakshaEval: An Indic Safety Benchmark for Multilingual LLMs
  2. 2026XL-SafetyBench: A Country-Grounded Cross-Cultural Benchmark for LLM Safety and Cultural Sensitivity
  3. 2025LinguaSafe: A Comprehensive Multilingual Safety Benchmark for Large Language Models
  4. 2026Safety Alignment Illusion: The Cross-Lingual Safety Gap in LLMs
  5. 2024RTP-LX: Can LLMs Evaluate Toxicity in Multilingual Scenarios?
  6. 2026BanglaVeilGuard: Cross-Script Safety Benchmarking and Lightweight Guardrails for Bangla Large Language Models
  7. 2026IndicSafeEval: Safety Robustness of Large Language Models under Multilingual Persuasive Jailbreak Attacks
  8. 2023All Languages Matter: On the Multilingual Safety of Large Language Models