Cover Image

SafeThinker: Reasoning about Risk to Deepen Safety Beyond Shallow Alignment

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Fang, Xianya, Luo, Xianying, Wang, Yadong, Chen, Xiang, Tian, Yu, Sun, Zequn, Liu, Rui, Fang, Jun, Tan, Naiqiang, Cui, Yuanning, Huang, Sheng-Jun
Format:	Preprint
Published:	2026
Subjects:	Cryptography and Security Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Compress the Easy, Explore the Hard: Difficulty-Aware Entropy Regularization for Efficient LLM Reasoning
by: Luo, Qin-Wen, et al.
Published: (2026)

Breaking the Reasoning Horizon in Entity Alignment Foundation Models
by: Cui, Yuanning, et al.
Published: (2026)

A Prompt-Based Knowledge Graph Foundation Model for Universal In-Context Reasoning
by: Cui, Yuanning, et al.
Published: (2024)

Beyond Superficial Unlearning: Sharpness-Aware Robust Erasure of Hallucinations in Multimodal LLMs
by: Fang, Xianya, et al.
Published: (2026)

Not All Thoughts are Generated Equal: Efficient LLM Reasoning via Multi-Turn Reinforcement Learning
by: Ning, Yansong, et al.
Published: (2025)

Agent-Omit: Adaptive Context Omission for Efficient LLM Agents
by: Ning, Yansong, et al.
Published: (2026)

KGFR: A Foundation Retriever for Generalized Knowledge Graph Question Answering
by: Cui, Yuanning, et al.
Published: (2025)

DeepTravel: An End-to-End Agentic Reinforcement Learning Framework for Autonomous Travel Planning Agents
by: Ning, Yansong, et al.
Published: (2025)

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

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

VeriThinker: Learning to Verify Makes Reasoning Model Efficient
by: Chen, Zigeng, et al.
Published: (2025)

TypedThinker: Diversify Large Language Model Reasoning with Typed Thinking
by: Wang, Danqing, et al.
Published: (2024)

DiMA: An LLM-Powered Ride-Hailing Assistant at DiDi
by: Ning, Yansong, et al.
Published: (2025)

SafeMLRM: Demystifying Safety in Multi-modal Large Reasoning Models
by: Fang, Junfeng, et al.
Published: (2025)

CameraBench: Benchmarking Visual Reasoning in MLLMs via Photography
by: Fang, I-Sheng, et al.
Published: (2025)

Bag of Tricks for Inference-time Computation of LLM Reasoning
by: Liu, Fan, et al.
Published: (2025)

Expanding the Scope: Inductive Knowledge Graph Reasoning with Multi-Starting Progressive Propagation
by: Shao, Zhoutian, et al.
Published: (2024)

Enhancing Reasoning Abilities of Small LLMs with Cognitive Alignment
by: Cai, Wenrui, et al.
Published: (2025)

SafeNeuron: Neuron-Level Safety Alignment for Large Language Models
by: Wang, Zhaoxin, et al.
Published: (2026)

Does Postgraduate Education Deepen Temporomandibular Disorders Insights for Dental Professionals?
by: Zejin Liu, et al.
Published: (2024)

Beyond Reactive Safety: Risk-Aware LLM Alignment via Long-Horizon Simulation
by: Sun, Chenkai, et al.
Published: (2025)

Beyond Dense States: Elevating Sparse Transcoders to Active Operators for Latent Reasoning
by: Wang, Yadong, et al.
Published: (2026)

Generating Explanations to Understand and Repair Embedding-based Entity Alignment
by: Tian, Xiaobin, et al.
Published: (2023)

ReThinker: Scientific Reasoning by Rethinking with Guided Reflection and Confidence Control
by: Tang, Zhentao, et al.
Published: (2026)

What Really Improves Mathematical Reasoning: Structured Reasoning Signals Beyond Pure Code
by: Zhao, Yuze, et al.
Published: (2026)

Unified Thinker: A General Reasoning Modular Core for Image Generation
by: Zhou, Sashuai, et al.
Published: (2026)

Deepening Divides
Published: (2020)

Molecular mechanism analyses of post‐traumatic epilepsy and hereditary epilepsy based on 10× single‐cell transcriptome sequencing technology
by: Fang Wen, et al.
Published: (2024)

SafeSteer: Localized On-Policy Distillation for Efficient Safety Alignment
by: Li, Hao, et al.
Published: (2026)

DDA-Thinker: Decoupled Dual-Atomic Reinforcement Learning for Reasoning-Driven Image Editing
by: Yang, Hanqing, et al.
Published: (2026)

Beyond Safe Answers: A Benchmark for Evaluating True Risk Awareness in Large Reasoning Models
by: Zheng, Baihui, et al.
Published: (2025)

The Missing Half: Unveiling Training-time Implicit Safety Risks Beyond Deployment
by: Zhang, Zhexin, et al.
Published: (2026)

KAG-Thinker: Interactive Thinking and Deep Reasoning in LLMs via Knowledge-Augmented Generation
by: Zhang, Dalong, et al.
Published: (2025)

DR-MMSearchAgent: Deepening Reasoning in Multimodal Search Agents
by: Wang, Shengqin, et al.
Published: (2026)

SafeWorld: Geo-Diverse Safety Alignment
by: Yin, Da, et al.
Published: (2024)

SafeAlign-VLA: A Negative-Enhanced Safe Alignment Framework for Risk-Aware Autonomous Driving
by: Tian, Kefei, et al.
Published: (2026)

The Association Between Socioeconomic Status and Cardiovascular Disease Risk in American Adults: Construction and Validation of a Nomogram Prediction Model Based on LASSO Feature Selection
by: Jun Li, et al.
Published: (2026)

SafeDrive: Fine-Grained Safety Reasoning for End-to-End Driving in a Sparse World
by: Kim, Jungho, et al.
Published: (2026)

The Thinker
Published: (2021)

ARM-Thinker: Reinforcing Multimodal Generative Reward Models with Agentic Tool Use and Visual Reasoning
by: Ding, Shengyuan, et al.
Published: (2025)