摘要：由于孟加拉语用户经常使用不同文字、拼写、代码混合形式和地区语体书写，采用以英语为中心或标准文字的基准，难以评估孟加拉语大语言模型（LLM）的安全性。本文提出 BanglaVeilGuard：一个紧凑型、孟加拉语优先的安全基准和轻量级提示词防护机制，覆盖六种语言形式：标准孟加拉语、罗马化孟加拉语、Banglish、孟加拉语--英语代码混合、噪声孟加拉语和方言孟加拉语。该基准包含 2,366 条经质量筛选的提示词，以及一个含 354 条提示词的独立评测划分，涵盖不安全、安全及安全敏感请求。BanglaVeilGuard 采用非破坏性的多视图规范化，并结合提示词风险分类器和生成前阈值门控机制，能够在不改变异构目标模型权重的情况下筛查提示词。在不同目标模型系列中，在确定性响应评分下，启用防护后的运行将 Claude Opus 4.8、BanglaLLama 和 TituLLM 的攻击成功率从 93.8--100.0% 降至 6.3%；搭载 BanglaVeilGuard 的 TigerLLM-1B 达到 78.2% 准确率和 8.8% 攻击成功率。该提示词防护机制还实现了 88.5% 的不安全请求召回率，显著高于所评估的仅提示词防护基线。其主要剩余代价是对方言和噪声形式的良性提示词过度拒答，这揭示了孟加拉语 LLM 部署中一项明确的安全性--有用性权衡边界。

Abstract:Bangla large language model (LLM) safety is difficult to evaluate with English-centric or standard-script benchmarks because Bangla users routinely write across scripts, spellings, code-mixed forms, and regional registers. This paper presents BanglaVeilGuard, a compact Bangla-first safety benchmark and lightweight prompt guard for six language forms: standard Bangla, Romanized Bangla, Banglish, code-mixed Bangla--English, noisy Bangla, and dialectal Bangla. The benchmark contains 2,366 quality-filtered prompts and a held-out 354-prompt evaluation split spanning unsafe, safe, and safe-sensitive requests. BanglaVeilGuard uses non-destructive multi-view normalization with a prompt-risk classifier and thresholded pre-generation gate, allowing it to screen prompts for heterogeneous target models without changing their weights. Across target-model families, guarded runs reduce attack success under deterministic response scoring from 93.8--100.0% to 6.3% for Claude Opus 4.8, BanglaLLama, and TituLLM; TigerLLM-1B with BanglaVeilGuard achieves 78.2% accuracy with 8.8% ASR. The prompt guard also attains 88.5% unsafe recall, substantially above the evaluated prompt-only guard baselines. The main remaining cost is over-refusal on dialectal and noisy benign prompts, revealing a concrete safety-helpfulness frontier for Bangla LLM deployment.