VLMGuard-R1：通过推理驱动的提示优化实现VLM的主动安全对齐
VLMGuard-R1: Proactive Safety Alignment for VLMs via Reasoning-Driven Prompt Optimization
AI总结：
本文提出VLMGuard-R1框架，通过多模态推理驱动的提示重写机制，利用三级推理流程训练重写器推断隐蔽威胁并生成定制响应，在不改变核心功能的前提下显著提升多种VLM架构的安全性。
摘要：
将视觉语言模型（VLM）与安全标准对齐，对于缓解其多模态复杂性带来的风险至关重要，因为融合视觉与语言会揭示传统安全防护无法触及的隐蔽威胁。受跨模态推理是预先应对复杂漏洞的关键这一洞察启发，我们提出了一种VLM安全的新方向：多模态推理驱动的提示重写。为此，我们提出了VLMGuard-R1，这是一个主动框架，通过推理引导的重写器来重写用户输入，动态解析文本-图像交互，从而在不改变核心功能的情况下提供增强安全性的提示，适用于各种VLM架构。为实现这一目标，我们设计了一个三级推理流程，用于合成一个数据集，以训练重写器推断隐蔽威胁，从而生成量身定制的可执行响应，而非通用拒绝。在五个VLM上的多个基准测试中的广泛实验表明，VLMGuard-R1优于四个基线。具体而言，VLMGuard-R1在SIUO基准上五个模型的平均安全性提升了41.59%。
英文摘要：
Aligning Vision-Language Models (VLMs) with safety standards is essential to mitigate risks arising from their multimodal complexity, where integrating vision and language unveils subtle threats beyond the reach of conventional safeguards. Inspired by the insight that reasoning across modalities is key to preempting intricate vulnerabilities, we propose a novel direction for VLM safety: multimodal reasoning-driven prompt rewriting. To this end, we introduce VLMGuard-R1, a proactive framework that refines user inputs through a reasoning-guided rewriter, dynamically interpreting text-image interactions to deliver refined prompts that bolster safety across diverse VLM architectures without altering their core parameters. To achieve this, we devise a three-stage reasoning pipeline to synthesize a dataset that trains the rewriter to infer subtle threats, enabling tailored, actionable responses over generic refusals. Extensive experiments across three benchmarks with five VLMs reveal that VLMGuard-R1 outperforms four baselines. In particular, VLMGuard-R1 achieves a remarkable 43.59\% increase in average safety across five models on the SIUO benchmark.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
