Think in Safety: Unveiling and Mitigating Safety Alignment Collapse in Multimodal Large Reasoning Model

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Lou, Xinyue, Li, You, Xu, Jinan, Shi, Xiangyu, Chen, Chi, Huang, Kaiyu
Format:	Preprint
Published:	2025
Subjects:	Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

When Helpers Become Hazards: A Benchmark for Analyzing Multimodal LLM-Powered Safety in Daily Life
by: Lou, Xinyue, et al.
Published: (2026)

TransportationGames: Benchmarking Transportation Knowledge of (Multimodal) Large Language Models
by: Zhang, Xue, et al.
Published: (2024)

Imagination Helps Visual Reasoning, But Not Yet in Latent Space
by: Li, You, et al.
Published: (2026)

SoT: Structured-of-Thought Prompting Guides Multilingual Reasoning in Large Language Models
by: Qi, Rui, et al.
Published: (2025)

Think Natively: Unlocking Multilingual Reasoning with Consistency-Enhanced Reinforcement Learning
by: Zhang, Xue, et al.
Published: (2025)

Migician: Revealing the Magic of Free-Form Multi-Image Grounding in Multimodal Large Language Models
by: Li, You, et al.
Published: (2025)

When Models Outthink Their Safety: Unveiling and Mitigating Self-Jailbreak in Large Reasoning Models
by: Mao, Yingzhi, et al.
Published: (2025)

MSR-Align: Policy-Grounded Multimodal Alignment for Safety-Aware Reasoning in Vision-Language Models
by: Xia, Yinan, et al.
Published: (2025)

Multilingual Collaborative Defense for Large Language Models
by: Li, Hongliang, et al.
Published: (2025)

Safety Alignment of Large Language Models via Contrasting Safe and Harmful Distributions
by: Zhang, Xiaoyun, et al.
Published: (2024)

Mitigating Fine-tuning based Jailbreak Attack with Backdoor Enhanced Safety Alignment
by: Wang, Jiongxiao, et al.
Published: (2024)

Internal Safety Collapse in Frontier Large Language Models
by: Wu, Yutao, et al.
Published: (2026)

DREAM: Disentangling Risks to Enhance Safety Alignment in Multimodal Large Language Models
by: Liu, Jianyu, et al.
Published: (2025)

Multi-Aspect Knowledge Distillation for Language Model with Low-rank Factorization
by: Liu, Zihe, et al.
Published: (2026)

VLSBench: Unveiling Visual Leakage in Multimodal Safety
by: Hu, Xuhao, et al.
Published: (2024)

STAIR: Improving Safety Alignment with Introspective Reasoning
by: Zhang, Yichi, et al.
Published: (2025)

Multimodal Cultural Safety: Evaluation Framework and Alignment Strategies
by: Qiu, Haoyi, et al.
Published: (2025)

Mitigating Exaggerated Safety in Large Language Models
by: Ray, Ruchira, et al.
Published: (2024)

Towards Comprehensive Post Safety Alignment of Large Language Models via Safety Patching
by: Zhao, Weixiang, et al.
Published: (2024)

Perception, Reason, Think, and Plan: A Survey on Large Multimodal Reasoning Models
by: Li, Yunxin, et al.
Published: (2025)

Towards Faster k-Nearest-Neighbor Machine Translation
by: Shi, Xiangyu, et al.
Published: (2023)

Safety in Large Reasoning Models: A Survey
by: Wang, Cheng, et al.
Published: (2025)

AdvChain: Adversarial Chain-of-Thought Tuning for Robust Safety Alignment of Large Reasoning Models
by: Zhu, Zihao, et al.
Published: (2025)

STAR-S: Improving Safety Alignment through Self-Taught Reasoning on Safety Rules
by: Wu, Di, et al.
Published: (2026)

R1-ACT: Efficient Reasoning Model Safety Alignment by Activating Safety Knowledge
by: In, Yeonjun, et al.
Published: (2025)

SafetyBench: Evaluating the Safety of Large Language Models
by: Zhang, Zhexin, et al.
Published: (2023)

Unforgotten Safety: Preserving Safety Alignment of Large Language Models with Continual Learning
by: Alssum, Lama, et al.
Published: (2025)

SAFER: Advancing Safety Alignment via Efficient Ex-Ante Reasoning
by: Feng, Kehua, et al.
Published: (2025)

SEA: Low-Resource Safety Alignment for Multimodal Large Language Models via Synthetic Embeddings
by: Lu, Weikai, et al.
Published: (2025)

MUCAR: Benchmarking Multilingual Cross-Modal Ambiguity Resolution for Multimodal Large Language Models
by: Wang, Xiaolong, et al.
Published: (2025)

Unveiling Trust in Multimodal Large Language Models: Evaluation, Analysis, and Mitigation
by: Zhang, Yichi, et al.
Published: (2025)

CM-Align: Consistency-based Multilingual Alignment for Large Language Models
by: Zhang, Xue, et al.
Published: (2025)

ReasoningGuard: Safeguarding Large Reasoning Models with Inference-time Safety Aha Moments
by: Wang, Yuquan, et al.
Published: (2025)

Think-Reflect-Revise: A Policy-Guided Reflective Framework for Safety Alignment in Large Vision Language Models
by: Weng, Fenghua, et al.
Published: (2025)

Unraveling and Mitigating Safety Alignment Degradation of Vision-Language Models
by: Liu, Qin, et al.
Published: (2024)

AM$^3$Safety: Towards Data Efficient Alignment of Multi-modal Multi-turn Safety for MLLMs
by: Zhu, Han, et al.
Published: (2026)

Circumventing Safety Alignment in Large Language Models Through Embedding Space Toxicity Attenuation
by: Zhang, Zhibo, et al.
Published: (2025)

EnchTable: Unified Safety Alignment Transfer in Fine-tuned Large Language Models
by: Wu, Jialin, et al.
Published: (2025)

Safety Alignment via Constrained Knowledge Unlearning
by: Shi, Zesheng, et al.
Published: (2025)

Guardians and Offenders: A Survey on Harmful Content Generation and Safety Mitigation of LLM
by: Zhang, Chi, et al.
Published: (2025)