REVEAL: Multi-turn Evaluation of Image-Input Harms for Vision LLM

• URL: http://arxiv.org/abs/2505.04673v1
• Date: Wed, 07 May 2025 10:09:55 GMT
• Title: REVEAL: Multi-turn Evaluation of Image-Input Harms for Vision LLM
• Authors: Madhur Jindal, Saurabh Deshpande,
• Abstract summary: We introduce the REVEAL Framework, a scalable and automated pipeline for evaluating image-input harms in Vision Large Language Models (VLLMs)<n>We extensively evaluated five state-of-the-art VLLMs, GPT-4o, Llama-3.2, Qwen2-VL, Phi3.5V, and Pixtral, across three important harm categories: sexual harm, violence, and misinformation.<n>GPT-4o demonstrated the most balanced performance as measured by our Safety-Usability Index (SUI) followed closely by Pixtral.
• Score: 0.098314893665023
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Vision Large Language Models (VLLMs) represent a significant advancement in artificial intelligence by integrating image-processing capabilities with textual understanding, thereby enhancing user interactions and expanding application domains. However, their increased complexity introduces novel safety and ethical challenges, particularly in multi-modal and multi-turn conversations. Traditional safety evaluation frameworks, designed for text-based, single-turn interactions, are inadequate for addressing these complexities. To bridge this gap, we introduce the REVEAL (Responsible Evaluation of Vision-Enabled AI LLMs) Framework, a scalable and automated pipeline for evaluating image-input harms in VLLMs. REVEAL includes automated image mining, synthetic adversarial data generation, multi-turn conversational expansion using crescendo attack strategies, and comprehensive harm assessment through evaluators like GPT-4o. We extensively evaluated five state-of-the-art VLLMs, GPT-4o, Llama-3.2, Qwen2-VL, Phi3.5V, and Pixtral, across three important harm categories: sexual harm, violence, and misinformation. Our findings reveal that multi-turn interactions result in significantly higher defect rates compared to single-turn evaluations, highlighting deeper vulnerabilities in VLLMs. Notably, GPT-4o demonstrated the most balanced performance as measured by our Safety-Usability Index (SUI) followed closely by Pixtral. Additionally, misinformation emerged as a critical area requiring enhanced contextual defenses. Llama-3.2 exhibited the highest MT defect rate ($16.55 \%$) while Qwen2-VL showed the highest MT refusal rate ($19.1 \%$).

Related papers

• Visual hallucination detection in large vision-language models via evidential conflict [24.465497252040294]
  Dempster-Shafer theory (DST)-based visual hallucination detection method for LVLMs through uncertainty estimation.<n>We propose to the best of our knowledge, the first Dempster-Shafer theory (DST)-based visual hallucination detection method for LVLMs through uncertainty estimation.
  arXiv  Detail & Related papers  (2025-06-24T11:03:10Z)
• Understanding and Benchmarking the Trustworthiness in Multimodal LLMs for Video Understanding [59.50808215134678]
  This study introduces Trust-videoLLMs, a first comprehensive benchmark evaluating 23 state-of-the-art videoLLMs.<n>Results reveal significant limitations in dynamic scene comprehension, cross-modal resilience and real-world risk mitigation.
  arXiv  Detail & Related papers  (2025-06-14T04:04:54Z)
• Prompt to Protection: A Comparative Study of Multimodal LLMs in Construction Hazard Recognition [0.0]
  This study conducts a comparative evaluation of five state-of-the-art large language models (LLMs)<n>Each model was tested under three prompting strategies: zero-shot, few-shot, and chain-of-thought (CoT)<n>Results reveal that CoT prompting significantly influenced performance, with CoT prompting consistently producing higher accuracy across models.
  arXiv  Detail & Related papers  (2025-06-09T05:22:35Z)
• VOILA: Evaluation of MLLMs For Perceptual Understanding and Analogical Reasoning [63.0285363282581]
  Multimodal Large Language Models (MLLMs) have become a powerful tool for integrating visual and textual information.<n>We introduce VOILA, a benchmark designed to evaluate MLLMs' perceptual understanding and abstract relational reasoning.<n>We reveal that current MLLMs struggle to comprehend inter-image relationships and exhibit limited capabilities in high-level relational reasoning.
  arXiv  Detail & Related papers  (2025-02-25T23:36:19Z)
• Calling a Spade a Heart: Gaslighting Multimodal Large Language Models via Negation [65.92001420372007]
  This paper systematically evaluates state-of-the-art MLLMs across diverse benchmarks.<n>We introduce the first benchmark GaslightingBench, specifically designed to evaluate the vulnerability of MLLMs to negation arguments.
  arXiv  Detail & Related papers  (2025-01-31T10:37:48Z)
• Retention Score: Quantifying Jailbreak Risks for Vision Language Models [60.48306899271866]
  Vision-Language Models (VLMs) are integrated with Large Language Models (LLMs) to enhance multi-modal machine learning capabilities.<n>This paper aims to assess the resilience of VLMs against jailbreak attacks that can compromise model safety compliance and result in harmful outputs.<n>To evaluate a VLM's ability to maintain its robustness against adversarial input perturbations, we propose a novel metric called the textbfRetention Score.
  arXiv  Detail & Related papers  (2024-12-23T13:05:51Z)
• Insight Over Sight? Exploring the Vision-Knowledge Conflicts in Multimodal LLMs [55.74117540987519]
  This paper explores the problem of commonsense-level vision-knowledge conflict in Multimodal Large Language Models (MLLMs) We introduce an automated pipeline, augmented with human-in-the-loop quality control, to establish a benchmark aimed at simulating and assessing the conflicts in MLLMs. We evaluate the conflict-resolution capabilities of nine representative MLLMs across various model families and find a noticeable over-reliance on textual queries.
  arXiv  Detail & Related papers  (2024-10-10T17:31:17Z)
• ETA: Evaluating Then Aligning Safety of Vision Language Models at Inference Time [12.160713548659457]
  adversarial visual inputs can easily bypass VLM defense mechanisms.<n>We propose a novel two-phase inference-time alignment framework, evaluating input visual contents and output responses.<n>Experiments show that ETA outperforms baseline methods in terms of harmlessness, helpfulness, and efficiency.
  arXiv  Detail & Related papers  (2024-10-09T07:21:43Z)
• DARE: Diverse Visual Question Answering with Robustness Evaluation [16.87867803628065]
  Vision Language Models (VLMs) extend remarkable capabilities of text-only large language models and vision-only models. They struggle with a number of crucial vision-language (VL) reasoning abilities such as counting and spatial reasoning. We introduce DARE, Diverse Visual Question Answering with Robustness Evaluation.
  arXiv  Detail & Related papers  (2024-09-26T16:31:50Z)
• How Many Unicorns Are in This Image? A Safety Evaluation Benchmark for Vision LLMs [55.91371032213854]
  This work focuses on the potential of Vision LLMs (VLLMs) in visual reasoning. We introduce a comprehensive safety evaluation suite, covering both out-of-distribution (OOD) generalization and adversarial robustness.
  arXiv  Detail & Related papers  (2023-11-27T18:59:42Z)
• On Evaluating Adversarial Robustness of Large Vision-Language Models [64.66104342002882]
  We evaluate the robustness of large vision-language models (VLMs) in the most realistic and high-risk setting. In particular, we first craft targeted adversarial examples against pretrained models such as CLIP and BLIP. Black-box queries on these VLMs can further improve the effectiveness of targeted evasion.
  arXiv  Detail & Related papers  (2023-05-26T13:49:44Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.