Imagen de Portada

Chain of Risk: Safety Failures in Large Reasoning Models and Mitigation via Adaptive Multi-Principle Steering

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Li, Xiaomin, Hou, Jianheng, Deng, Zheyuan, Zhang, Zhiwei, Li, Taoran, Lu, Binghang, Hu, Bing, Zhao, Yunhan, Hao, Yuexing
Formato:	Preprint
Publicado:	2026
Materias:	Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Muon-OGD: Muon-based Spectral Orthogonal Gradient Projection for LLM Continual Learning
por: Lu, Binghang, et al.
Publicado: (2026)

When Think-with-Image Meets Safety: What Determines Multimodal Jailbreak Robustness?
por: Tian, Yuan, et al.
Publicado: (2026)

MedGUIDE: Benchmarking Clinical Decision-Making in Large Language Models
por: Li, Xiaomin, et al.
Publicado: (2025)

Dual Identities, Singular Journeys: How Working Tourists' Role Perception Shapes Place Attachment Through Memorable Tourism Experience
por: Yelin Fang, et al.
Publicado: (2026)

Steering Multimodal Large Language Models Decoding for Context-Aware Safety
por: Liu, Zheyuan, et al.
Publicado: (2025)

FAAGC: Feature Augmentation on Adaptive Geodesic Curve Based on the shape space theory
por: Han, Yuexing, et al.
Publicado: (2025)

AdamFLIP: Adaptive Momentum Feedback Linearization Optimization for Hard Constrained PINN Training
por: Lu, Binghang, et al.
Publicado: (2026)

Dynamics of Apparent Horizon and a Null Comparison Principle
por: An, Xinliang, et al.
Publicado: (2023)

ProbeLLM: Automating Principled Diagnosis of LLM Failures
por: Huang, Yue, et al.
Publicado: (2026)

Attention Misses Visual Risk: Risk-Adaptive Steering for Multimodal Safety Alignment
por: Park, Jonghyun, et al.
Publicado: (2025)

Mitigating Overthinking in Large Reasoning Models via Manifold Steering
por: Huang, Yao, et al.
Publicado: (2025)

Data-adaptive Safety Rules for Training Reward Models
por: Li, Xiaomin, et al.
Publicado: (2025)

Chain of Mindset: Reasoning with Adaptive Cognitive Modes
por: Jiang, Tianyi, et al.
Publicado: (2026)

Temperature and wind characteristics of Lenghu site for ventilation and structural design of large telescope enclosure
por: Li, Taoran, et al.
Publicado: (2025)

Mitigating Safety Tax via Distribution-Grounded Refinement in Large Reasoning Models
por: Xie, Yingsha, et al.
Publicado: (2026)

Mitigating Overthinking in Large Reasoning Language Models via Reasoning Path Deviation Monitoring
por: Guan, Weixin, et al.
Publicado: (2026)

Adaptive Residual-Update Steering for Low-Overhead Hallucination Mitigation in Large Vision Language Models
por: Zou, Zhengtao, et al.
Publicado: (2025)

Dot1l Regulates the Spontaneous Bone Regeneration of Periosteum‐Derived Stem Cells by Regulating Chac1 Expression
por: Taoran Jiang, et al.
Publicado: (2025)

fPINN-DeepONet: A Physics-Informed Operator Learning Framework for Multi-term Time-fractional Mixed Diffusion-wave Equations
por: Lu, Binghang, et al.
Publicado: (2026)

Think in Safety: Unveiling and Mitigating Safety Alignment Collapse in Multimodal Large Reasoning Model
por: Lou, Xinyue, et al.
Publicado: (2025)

Answering the Unanswerable Is to Err Knowingly: Analyzing and Mitigating Abstention Failures in Large Reasoning Models
por: Liu, Yi, et al.
Publicado: (2025)

Muon with Spectral Guidance: Efficient Optimization for Scientific Machine Learning
por: Lu, Binghang, et al.
Publicado: (2026)

iPINNER: An Iterative Physics-Informed Neural Network with Ensemble Kalman Filter
por: Lu, Binghang, et al.
Publicado: (2025)

Morephy-Net: An Evolutionary Multi-objective Optimization for Replica-Exchange-based Physics-informed Neural Operator Learning Networks
por: Lu, Binghang, et al.
Publicado: (2025)

Neural-POD: A Plug-and-Play Neural Operator Framework for Infinite-Dimensional Functional Nonlinear Proper Orthogonal Decomposition
por: Mou, Changhong, et al.
Publicado: (2026)

Investigating the Effectiveness of a Socratic Chain-of-Thoughts Reasoning Method for Task Planning in Robotics, A Case Study
por: Bot, Veronica, et al.
Publicado: (2025)

Understanding mechanisms underlying solar cycle predictability with a general framework
por: Li, Binghang, et al.
Publicado: (2026)

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

Mitigating Cognitive Inertia in Large Reasoning Models via Latent Spike Steering
por: Lee, Seojin, et al.
Publicado: (2026)

VEGAS: Mitigating Hallucinations in Large Vision-Language Models via Vision-Encoder Attention Guided Adaptive Steering
por: Wang, Zihu, et al.
Publicado: (2025)

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

FAGC:Feature Augmentation on Geodesic Curve in the Pre-Shape Space
por: Han, Yuexing, et al.
Publicado: (2023)

Risk Management for Mitigating Benchmark Failure Modes: BenchRisk
por: McGregor, Sean, et al.
Publicado: (2025)

SCANS: Mitigating the Exaggerated Safety for LLMs via Safety-Conscious Activation Steering
por: Cao, Zouying, et al.
Publicado: (2024)

Feature Extraction and Steering for Enhanced Chain-of-Thought Reasoning in Language Models
por: Li, Zihao, et al.
Publicado: (2025)

Advanced Electrolyte Solution for Aqueous Lithium‐Ion Batteries: Extending Their Lifespan
por: Binghang Liu, et al.
Publicado: (2025)

Beware of Your Po! Measuring and Mitigating AI Safety Risks in Role-Play Fine-Tuning of LLMs
por: Zhao, Weixiang, et al.
Publicado: (2025)

FAGStyle: Feature Augmentation on Geodesic Surface for Zero-shot Text-guided Diffusion Image Style Transfer
por: Han, Yuexing, et al.
Publicado: (2024)

Few-shot Image Generation via Information Transfer from the Built Geodesic Surface
por: Han, Yuexing, et al.
Publicado: (2024)

Lifting the Veil on Composition, Risks, and Mitigations of the Large Language Model Supply Chain
por: Huang, Kaifeng, et al.
Publicado: (2024)