Vision-Language Models (VLMs) extend Large Language Models (LLMs) with visual perception capabilities, unlocking broad applications across many domains. However, ensuring their safety remains a critical challenge, as adversarial visual inputs can easily bypass built-in safeguards and elicit harmful content. In this paper, we uncover a phenomenon we call delayed safety awareness, where a jailbroken VLM initially produces harmful content but ultimately recognizes the harmfulness at the end of the generation process. We attribute this phenomenon to the fact that the model's safety awareness against jailbreaks cannot be effectively transferred to the intermediate stages of text generation. Motivated by this insight, we introduce SafetyReminder, a simple yet effective defense that optimizes a learnable soft prompt using our proposed Safety-Activation Prompt Tuning (SAPT). This soft prompt is inserted into the generated text to activate the safety awareness of the model, steering it toward refusal when harmful content arises while preserving helpfulness in benign scenarios. We evaluate our method on three established harmful benchmarks and across three types of adversarial attacks. Experimental results demonstrate that our method achieves state-of-the-art defense performance with strong generalization, offering a practical and lightweight solution for safe deployment of VLMs.

Collecting and annotating data for semantic segmentation can end up costing a lot of time and energy. Unsupervised Domain Adaptation (UDA) for semantic segmentation allows models trained on certain source domain data (such as the GTA synthetic dataset) to be applied to certain target data (like the Cityscapes real dataset), significantly decreasing the requirement for target-domain manual pixel-level annotations.In this paper, we suggest a cross-domain unified semantic segmentation network training framework, the Attention Distribution Adaptation Network (ADAN). It (1) proposes the Content Consistency Transformation (CCT), which maps source domain data to an intermediate domain that has a data distribution that is comparable to that of the target domain, and (2) introduces the Attention Adaptation Module (AAM), which bolsters the synchronization of attention between the intermediate domain and the target domain. ADAN achieved unprecedented mIoUs of 77.2 and 68.8 on GTA→Cityscapes and Synthia→Cityscapes, respectively, corresponding to improvements of +1.3 and +0.6 over the state of the art.

Motion planning is a difficult task, especially when generating feasible future trajectories in complex and interactive scenarios. While recent advancements in imitation-based planning have shown significant progress, this approach often encounters causal confusion in dynamic traffic environments. This confusion will cause the planner to incorrectly associate certain actions with outcomes, leading to suboptimal or unsafe plans. To address this, we introduce a novel framework called$\overline{C}^{2}L$, which improves the planner's latent Causal understanding by incorporating Contrastive Learning and counterfactual data augmentation. Additionally, we propose a shortcut eliminator to extract copycat-free features from history states, reducing the impact of temporal spurious correlations. We validate our method on the nuPlan and interPlan benchmarks, with extensive experiments demonstrating that$C^{2}L$delivers highly competitive performance compared to state-of-the-art methods.