Safety-aligned Large Language Models (LLMs) still show two dominant failure modes: they are easily jailbroken, or they over-refuse harmless inputs that contain sensitive surface signals. We trace both to a common cause: current models reason weakly about links between actions and outcomes and over-rely on surface-form signals, lexical or stylistic cues that do not encode consequences. We define this failure mode as Consequence-blindness. To study consequence-blindness, we build a benchmark named CB-Bench covering four risk scenarios that vary whether semantic risk aligns with outcome risk, enabling evaluation under both matched and mismatched conditions which are often ignored by existing safety benchmarks. Mainstream models consistently fail to separate these risks and exhibit consequence-blindness, indicating that consequence-blindness is widespread and systematic. To mitigate consequence-blindness, we introduce CS-Chain-4k, a consequence-reasoning dataset for safety alignment. Models fine-tuned on CS-Chain-4k show clear gains against semantic-camouflage jailbreaks and reduce over-refusal on harmless inputs, while maintaining utility and generalization on other benchmarks. These results clarify the limits of current alignment, establish consequence-aware reasoning as a core alignment goal and provide a more practical and reproducible evaluation path.
Exploring Adversarial Robustness and Safety Alignment in Multilingual Multi-Modal Large Language Models
Mohamed Bin Zayed University of AI, UAE(穆罕默德·本·扎耶德人工智能大学,阿联酋);Khalifa University, UAE(卡比拉大学,阿联酋);Australian National University, Australia(澳大利亚国立大学,澳大利亚)
InSaAF: Incorporating Safety through Accuracy and Fairness | Are LLMs ready for the Indian Legal Domain?
Centre for Responsible AI, Indian Institute of Technology Madras(印度马德拉斯理工学院负责任人工智能中心);International Institute of Information Technology, Hyderabad(海德拉巴国际信息技术学院);AmexAI Labs, American Express(美国运通AmexAI实验室);Wadhwani School of Data Science and AI, Indian Institute of Technology Madras(印度马德拉斯理工学院瓦德瓦尼数据科学与人工智能学院)
AdvChain: Adversarial Chain-of-Thought Tuning for Robust Safety Alignment of Large Reasoning Models
The Chinese University of Hong Kong, Shenzhen(香港中文大学(深圳));State University of New York at Buffalo(纽约州立大学布法罗分校);Huawei International, Singapore(新加坡华为国际)
SurakshaEval: An Indic Safety Benchmark for Multilingual LLMs
Inception42;Mohamed bin Zayed University of Artificial Intelligence(穆罕默德·本·扎耶德人工智能大学);University of Michigan(密歇根大学);Institute of Computer Science Artificial Intelligence & Technology(计算机科学人工智能与技术研究院);International Institute of Information Technology Hyderabad(海得拉巴国际信息技术学院);Ministry of Electronics and IT India(印度电子与信息技术部);Indian Institute of Technology Kharagpur(印度理工学院卡拉格普尔分校)
Manacá-1B-Instruct-2: Reproducible Safety Alignment for a Native Brazilian-Portuguese LLM. Negative DPO Results, a Controlled Composition Scan, and a Silent Failure Mode of the LLM Judge
National Laboratory for Scientific Computing (LNCC)(国家科学计算实验室);Institute of Artificial Intelligence(人工智能研究所)
Are Vision LLMs Road-Ready? A Comprehensive Benchmark for Safety-Critical Driving Video Understanding
Department of Computer Science, Virginia Tech(弗吉尼亚理工学院计算机科学系);Virginia Tech Transportation Institute, Virginia Tech(弗吉尼亚理工学院交通研究所);Department of Statistics, Virginia Tech(弗吉尼亚理工学院统计学系)
SceneWeave: Audience-Guided Scene Exploration for Dramatic Adaptation
The Central Academy of Drama(中央戏剧学院);Hithink RoyalFlush Information Network Co., Ltd.(同花顺信息技术股份有限公司);China Telecom Artificial Intelligence Technology (Beijing) Co., Ltd.(中国电信人工智能技术(北京)有限公司);Beijing Institute of Mathematical Sciences and Applications (BIMSA)(北京数学科学与应用研究院)