Multimodal Multi-Turn Safety Alignment

Latest papers

10 papers1 paper in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.

When Reasoning Is Not Enough: Proactive Inquiry for Cross-Modal Safety in Multimodal Large Language Models
Oct 15, 2026Wei Cai, Jian Zhao, Yuchen Yuan +1Multimodal Multi-Turn Safety Alignment VLM Safety
Institute of Artificial Intelligence (TeleAI), China Telecom, P. R. China. · Peking University. · School of Artificial Intelligence, OPtics and ElectroNics (iOPEN), Northwestern Polytechnical University, Xi’an 710072, P. R. China.
ShieldCLIP: Selective Safety Alignment for Harmful Content Mitigation in Multimodal Foundation Models
Sep 30, 2026Tobia Poppi, Silvia Cappelletti, Samuele Poppi +4Cross-Modal Alignment Multimodal Multi-Turn Safety Alignment
University of Modena and Reggio Emilia, Modena, Italy. · University of Pisa, Pisa, Italy. · MBZUAI, Abu Dhabi, United Arab Emirates. +1
Transfer Safety Awareness for Cross-Modal Safety Drift in Multimodal Large Language Models
Sep 2, 2026Tianqi Xiao, Shiyao Cui, Minghao Zhang +2Multimodal Multi-Turn Safety Alignment Multimodal Large Language Models
1The Conversational AI (CoAI) group, DCST, Tsinghua University · 3Northwestern Polytechnical University · School of Cyberspace Security, Beijing University of Posts and Telecommunications +1
SafeCap: Improving LVLM Safety with Image Captioning Reinforcement Learning
Aug 11, 2026Caoyuan Ma, Wenpu Liu, Weichu Xie +12VLM Robustness Adversarial Attacks on VLMs
The University of Tokyo · Peking University · Wuhan University +4
SafeNexus: Discovering and Steering Modality-Universal Safety Neurons in MLLMs
Jul 31, 2026Jian Yu, Fei Shen, Cong Wang +5Multimodal Robustness Multimodal Large Language Models
1Nanjing University of Science and Technology · 2National University of Singapore · 3Nanjing University +2
Harnessing Textual Refusal Directions for Multimodal Safety
Jun 30, 2026Moreno D'Incà, Nicu Sebe, Massimiliano ManciniLanguage Model Safety Evaluation Multimodal Large Language Models
University of Trento
SaFeR-Steer: Evolving Multi-Turn MLLMs via Synthetic Bootstrapping and Feedback Dynamics
Mar 18, 2026Haolong Hu, Hanyu Li, Tiancheng He +6Language Model Steering Multimodal Large Language Models
Huazhong University of Science and Technology · Beijing University of Posts and Telecommunications · West China Biomedical Big Data Center, Sichuan University +2
Visual Distraction Undermines Moral Reasoning in Vision-Language Models
Mar 17, 2026Xinyi Yang, Chenheng Xu, Weijun Hong +4Multimodal Multi-Turn Safety Alignment VLM Reasoning
Institute for Artificial Intelligence, Peking University · School of Psychological and Cognitive Sciences, Peking University · State Key Lab of General Artificial Intelligence, Peking University +3
Co-Evolutionary Multi-Modal Alignment via Structured Adversarial Evolution
Mar 2, 2026Guoxin Shi, Haoyu Wang, Zaihui Yang +2Multimodal Multi-Turn Safety Alignment Adversarial Training
Nanyang Technological University · Tsinghua University
Towards Multi-modal Multi-turn Safety: From Agentic Interaction to Strategic Alignment
Jan 8, 2026Han Zhu, Jiale Chen, Chengkun Cai +8Language Model Safety Evaluation Multimodal Large Language Models
Hong Kong University of Science and Technology · University College London · AISpeech +1

Oct 15, 2026J/K move · Enter open · S save

When Reasoning Is Not Enough: Proactive Inquiry for Cross-Modal Safety in Multimodal Large Language Models

Wei Cai, Jian Zhao, Yuchen Yuan, Xuelong Li

Institute of Artificial Intelligence (TeleAI), China Telecom, P. R. China. · Peking University. · School of Artificial Intelligence, OPtics and ElectroNics (iOPEN), Northwestern Polytechnical University, Xi’an 710072, P. R. China.

Multimodal Multi-Turn Safety Alignment VLM Safety LLM Safety MLLM Safety Multimodal Large Language Models Clarification Question Generation+1

Open paper AskPDF arXiv

Multimodal large language models (MLLMs) can act unsafely when individually benign images and texts are interpreted jointly. Beyond implicit reasoning risk, where reasoning drifts toward harm, we identify intent-level risk, where cross-modal ambiguity silently commits the model to a hazardous premise that path-level optimization alone cannot correct. We propose SAFE-PRO, which uses a Risk-Conditioned Uncertainty (RCU) signal, combining predictive entropy with a learned cross-modal risk score, to decide when to ask safety-grounded clarifying questions and where to apply path optimization, under a reward that prioritizes safety and penalizes over-refusal. We also build SSUI-Dialog, a multimodal multi-turn dataset with dual-intent supervision, and theoretically analyze gradient variance and unsafe-completion risk under RCU. Across seven benchmarks and two base MLLMs, SAFE-PRO consistently improves safety over SRPO, roughly halves its over-refusal, and surpasses GPT-4o and Claude-3.5-Sonnet in average single-turn safety.