OOD-MMSafe: 推进多模态大语言模型安全性从有害意图到隐藏后果
OOD-MMSafe: Advancing MLLM Safety from Harmful Intent to Hidden Consequences
AI总结：
OOD-MMSafe通过CASPO框架提升多模态大语言模型对隐藏后果的识别能力，显著降低风险识别失败率。
摘要：
尽管多模态大语言模型（MLLMs）的安全对齐已经引起了广泛关注，但当前的范式主要针对恶意意图或情境违规。我们提出将安全前沿转向后果驱动的安全，这是自主和具身代理稳健部署的关键范式。为了正式实现这一转变，我们引入OOD-MMSafe，一个包含455个精心挑选的查询-图像对的基准，旨在评估模型识别上下文依赖因果链中潜在危害的能力。我们的分析揭示了前沿模型普遍存在因果盲区，最高67.5%的失败率出现在高容量闭源模型中，并识别出一个偏好上限，其中静态对齐导致格式中心的失败而非安全推理的改进。为了解决这些瓶颈，我们开发了后果意识安全策略优化（CASPO）框架，该框架将模型的内在推理作为动态参考用于token级自我蒸馏奖励。实验结果表明，CASPO显著增强了后果投影，使风险识别的失败率降至7.3%（Qwen2.5-VL-7B）和5.7%（Qwen3-VL-4B），同时保持整体有效性。
英文摘要：
While safety alignment for Multimodal Large Language Models (MLLMs) has gained significant attention, current paradigms primarily target malicious intent or situational violations. We propose shifting the safety frontier toward consequence-driven safety, a paradigm essential for the robust deployment of autonomous and embodied agents. To formalize this shift, we introduce OOD-MMSafe, a benchmark comprising 455 curated query-image pairs designed to evaluate a model's ability to identify latent hazards within context-dependent causal chains. Our analysis reveals a pervasive causal blindness among frontier models, with the highest 67.5% failure rate in high-capacity closed-source models, and identifies a preference ceiling where static alignment yields format-centric failures rather than improved safety reasoning as model capacity grows. To address these bottlenecks, we develop the Consequence-Aware Safety Policy Optimization (CASPO) framework, which integrates the model's intrinsic reasoning as a dynamic reference for token-level self-distillation rewards. Experimental results demonstrate that CASPO significantly enhances consequence projection, reducing the failure ratio of risk identification to 7.3% for Qwen2.5-VL-7B and 5.7% for Qwen3-VL-4B while maintaining overall effectiveness.
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。
