SafeAtlas-VL: Beyond Binary Multimodal Safety with Large-Scale Data and Guard Models

• 类型：arxiv
• 标识：2608.29098
• 链接：https://arxiv.org/abs/2608.29098
• 主分类：risk
• 形态：method
• TLDR：Multimodal safety moderation requires distinguishing risks arising from visual content, user intent, and assistant behavior. Existing safeguards, however, are typically trained for a single judgment target and reduce safety assessment to a binary decision. Consequently, risk becomes difficult to compare across a multimodal interaction, and ambiguous cases are obscured. We introduce SafeAtlas-VL, a dataset of 1.5M training instances that places image-, request-, and response-level judgments on a five-level ordered scale. We curate a broad collection of safety-relevant data from both real-world
• 副分类：multimodal
• 待LLM分类：否
• 标题中文：SafeAtlas-VL：超越二元判断的大规模多模态安全数据与防护模型
• TLDR中文：多模态安全审核需要区分源自视觉内容、用户意图与助手行为的风险。然而现有防护模型通常只针对单一判断目标训练，将安全评估简化为二元决策，导致风险难以在多模态交互中跨类型比较，模糊案例也被掩盖。我们提出 SafeAtlas-VL，一个包含 150 万训练实例的数据集，将图像、请求、回答三个层级判断统一在五级有序量表上，并从真实世界
• 来源文件：
• /inbox/tom/_candidates/2026-09-01-agent-rag-longcontext-candidates.json