Chain-of-Thought (CoT) prompting is widely adopted for mathematical problem solving, including in low-resource languages, yet its robustness under irrelevant context - a known vulnerability in English - remains unexamined in such settings. To systematically probe this challenge, we introduce DISTRACTMATH-BN, a Bangla benchmark that augments MGSM and MSVAMP with semantically coherent but computationally irrelevant information. Evaluating seven models ranging from 3B to 12B parameters, we observe substantial performance degradation under distractors: standard models drop up to 41 points, while reasoning-specialized models decline 14-20 points despite consuming 5× more tokens. We propose †DAGGER, which reformulates mathematical problem solving as executable computational graph generation with dedicated modeling of distractor nodes. Fine-tuning Gemma-3 models using Supervised Fine-Tuning followed by Group Relative Policy Optimization achieves comparable weighted accuracy on augmented benchmarks while using 89% fewer tokens than reasoning models. Importantly, this robustness emerges without explicit training on distractor-augmented examples. Our results suggest that enforcing structured intermediate representations can improve robustness and inference efficiency in mathematical understanding compared to free-form reasoning, particularly in noisy, low-resource settings.

Vision Language Models (VLMs) are increasingly integrated into privacy-critical domains, yet existing evaluations of personally identifiable information (PII) leakage largely treat privacy as a static extraction task and ignore how a subject’s online presence—the volume of their data available online—influences privacy alignment. We introduce PII-VisBench, a novel benchmark containing 4,000 unique probes designed to evaluate VLM safety through the continuum of online presence. The benchmark stratifies 200 subjects into four visibility categories: high, medium, low, and zero—based on the extent and nature of their information available online. We evaluate 18 open-source VLMs (0.3B–32B) based on two key metrics: percentage of PII probing queries refused (Refusal Rate) and the fraction of non-refusal responses flagged for containing PII (Conditional PII Disclosure Rate). Across models, we observe a consistent pattern: refusals increase and PII disclosures decrease (9.10% high → 5.34% low) as subject visibility drops. We identify that models are more likely to disclose PII for high-visibility subjects, alongside substantial model-family heterogeneity and PII-type disparities. Finally, paraphrasing and jailbreak-style prompts expose attack- and model-dependent failures, motivating visibility-aware safety evaluation and training interventions.