mlsanityData to 7 Oct 2026

Better Safe Than Sorry? Overreaction Problem of Vision Language Models in Visual Emergency Recognition

arXiv 2505.15367 ↗First posted 21 May 2025cs.CV cs.AI cs.CLcited by 7

On the map: Multimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness

Abstract (arXiv)

Vision-Language Models (VLMs) have shown capabilities in interpreting visual content, but their reliability in safety-critical scenarios remains insufficiently explored. We introduce VERI, a diagnostic benchmark comprising 200 synthetic images (100 contrastive pairs) and an additional 50 real-world images (25 pairs) for validation. Each emergency scene is paired with a visually similar but safe counterpart through human verification. Using a two-stage evaluation protocol (risk identification and emergency response), we assess 17 VLMs across medical emergencies, accidents, and natural disasters. Our analysis reveals an "overreaction problem": models achieve high recall (70-100%) but suffer from low precision, misclassifying 31-96% of safe situations as dangerous. Seven safe scenarios were universally misclassified by all models. This "better-safe-than-sorry" bias stems from contextual overinterpretation (88-98% of errors). Both synthetic and real-world datasets confirm these systematic patterns, challenging VLM reliability in safety-critical applications. Addressing this requires enhanced contextual reasoning in ambiguous visual situations.

Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Placed in MLLM jailbreaks and adversarial robustness, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 18 of its 34 references that are in the map. First posted on arXiv 21 May 2025.

Builds on

18 of 34 references.

References · 18

  1. 2020Scaling Laws for Neural Language Modelscited by 6,411
  2. 2025Qwen2.5-VL Technical Reportcited by 5,418
  3. 2019nuScenes: A multimodal dataset for autonomous drivingcited by 4,485
  4. 2024GPT-4o System Cardcited by 4,467
  5. 2024Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phonecited by 1,883
  6. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,769
  7. 2023Vision-Language Models for Vision Tasks: A Surveycited by 834
  8. 2024What matters when building vision-language models?cited by 321
  9. 2024Pixtral 12Bcited by 174
  10. 2023How Many Unicorns Are in This Image? A Safety Evaluation Benchmark for Vision LLMscited by 103
  11. 2024Examining Gender and Racial Bias in Large Vision-Language Models Using a Novel Dataset of Parallel Imagescited by 55
  12. 2024ScVLM: Enhancing Vision-Language Model for Safety-Critical Event Understandingcited by 13
6 more
  1. 2022Improving Commonsense in Vision-Language Models via Knowledge Graph Riddlescited by 7
  2. 2024ERVQA: A Dataset to Benchmark the Readiness of Large Vision Language Models in Hospital Environmentscited by 5
  3. 2025Leveraging Large Language Models for Enhanced Digital Twin Modeling: Trends, Methods, and Challengescited by 4
  4. 2025Evaluation of Safety Cognition Capability in Vision-Language Models for Autonomous Drivingcited by 4
  5. 2025"What If Smart Homes Could See Our Homes?": Exploring DIY Smart Home Building Experiences with VLM-Based Camera Sensorscited by 3
  6. 2025Balancing Complexity and Informativeness in LLM-Based Clustering: Finding the Goldilocks Zonecited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 1; 2026: 6.

7 papers.

  1. 2025Jailbreaking on Text-to-Video Models via Scene Splitting Strategycited by 10
  2. 2026When Context Flips, Safety Breaks: Diagnosing Brittle Safety in Aligned Language Modelscited by 2
  3. 2026SaFeR-ToolKit: Structured Reasoning via Virtual Tool Calling for Multimodal Safetycited by 0
  4. 2026EgoSafetyBench: A Diagnostic Egocentric Video Benchmark for Evaluating Embodied VLMs as Runtime Safety Guardscited by 0
  5. 2026Autonomous VR-Based Risk Detection for Situational Awareness in Dangerous Settingscited by 0
  6. 2026Hazard or Anomaly? Evaluating VLMs for Understanding Dangers and Discrepanciescited by 0
  7. 2026Visual Compliance via Executable Safety Rule Entailmentcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026EviSafe: Evidence-Grounded Safety Evaluation for Vision-Language Models
  2. 2025Rethinking Bottlenecks in Safety Fine-Tuning of Vision Language Models
  3. 2025VSCBench: Bridging the Gap in Vision-Language Model Safety Calibration
  4. 2026From Local Evidence to Safety Verdicts: Causal Tracing in Vision-Language Models
  5. 2025VLSU: Mapping the Limits of Joint Multimodal Understanding for AI Safety
  6. 2025DUAL-Bench: Measuring Over-Refusal and Robustness in Vision-Language Models
  7. 2026Consistent but Dangerous: Per-Sample Safety Classification Reveals False Reliability in Medical Vision-Language Models
  8. 2026HorusEye: Language as Dynamic Attention for Emergency Visual Analysis