利用文本拒绝方向实现多模态安全
Harnessing Textual Refusal Directions for Multimodal Safety
AI总结：
提出MARS方法，通过从LLM骨干提取的文本拒绝方向泛化到多模态，无需多模态安全数据即可提升安全性，在五个先进MLLM上验证了有效性。
摘要：
为了提高大型语言模型（LLM）的安全性，我们可以进行训练后对齐或利用激活空间中的拒绝方向。这两种策略在多模态LLM（MLLM）中不太可行，因为它们需要不安全的多模态数据，而这类数据比单模态数据更难收集。在这项工作中，我们放宽了这一约束，并研究了直接从LLM骨干中提取的文本拒绝方向是否能够跨模态（如图像、视频）泛化。初步发现证实了这种能力，尽管有效性受层选择、引导强度和跨模态对齐的影响，其中跨模态对齐会导致安全的多模态输入被错误地引导至拒绝。在此基础上，我们引入了模态无关拒绝引导（MARS），这是一种轻量级、无需训练的方法，无需多模态安全数据即可注入多模态安全性。MARS通过激活重中心化纠正模态错位，在几何定义的信任区域内自适应地缩放引导强度，并选择最优干预层，在生成的第一个token上操作。在五个最先进的MLLM上，针对安全性、实用性和视频越狱基准进行评估，MARS在保持实用性的同时实现了持续的安全性提升。这些结果表明，安全相关结构在模态间共享，文本拒绝方向是多模态对齐的一个强大且未被充分探索的基础。
英文摘要：
To improve safety in Large Language Models (LLMs) we can either perform post-training alignment or exploit refusal directions in the activation space. Both strategies are less feasible in Multimodal LLMs (MLLMs) as they require unsafe multimodal data, harder to collect than their unimodal counterpart. In this work, we relax this constraint and investigate whether textual refusal directions, extracted directly from the LLM backbone, generalize across modalities (i.e., image, video). Preliminary findings confirm this ability, though effectiveness is conditioned by layer selection, steering strength, and cross-modal alignment, with the latter causing safe multimodal inputs to be spuriously steered toward refusal. Building on this, we introduce Modality-Agnostic Refusal Steering (MARS), a light-weight training-free approach that injects multimodal safety without the need for multimodal safety data. MARS corrects modality misalignment via activation re-centering, adaptively scales steering strength within a geometrically defined trust region, and selects the optimal intervention layer, operating at the first generated token. Evaluated on five SOTA MLLMs across safety, utility, and video jailbreak benchmarks, MARS achieves consistent safety gains while preserving utility. These results reveal that safety-relevant structure is shared across modalities and that textual refusal directions are a powerful and underexplored foundation for multimodal alignment.
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。
