Cover Image

The Devil behind the mask: An emergent safety vulnerability of Diffusion LLMs

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wen, Zichen, Qu, Jiashu, Chen, Zhaorun, Lu, Xiaoya, Liu, Dongrui, Liu, Zhiyuan, Wu, Ruixi, Yang, Yicun, Jin, Xiangqi, Xu, Haoyun, Liu, Xuyang, Li, Weijia, Lu, Chaochao, Shao, Jing, He, Conghui, Zhang, Linfeng
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

Thinking Inside the Mask: In-Place Prompting in Diffusion LLMs
by: Jin, Xiangqi, et al.
Published: (2025)

Dataset Distillation with Neural Characteristic Function: A Minmax Perspective
by: Wang, Shaobo, et al.
Published: (2025)

X-Boundary: Establishing Exact Safety Boundary to Shield LLMs from Multi-Turn Jailbreaks without Compromising Usability
by: Lu, Xiaoya, et al.
Published: (2025)

INFA-Guard: Mitigating Malicious Propagation via Infection-Aware Safeguarding in LLM-Based Multi-Agent Systems
by: Zhou, Yijin, et al.
Published: (2026)

Respecting Self-Uncertainty in On-Policy Self-Distillation for Efficient LLM Reasoning
by: Ke, Junlong, et al.
Published: (2026)

Token Pruning in Multimodal Large Language Models: Are We Solving the Right Problem?
by: Wen, Zichen, et al.
Published: (2025)

Towards Principled Dataset Distillation: A Spectral Distribution Perspective
by: Wu, Ruixi, et al.
Published: (2026)

IPCV: Information-Preserving Compression for MLLM Visual Encoders
by: Chen, Yuan, et al.
Published: (2025)

LLMs know their vulnerabilities: Uncover Safety Gaps through Natural Distribution Shifts
by: Ren, Qibing, et al.
Published: (2024)

LLMs Deceive Unintentionally: Emergent Misalignment in Dishonesty from Misaligned Samples to Biased Human-AI Interactions
by: Hu, Xuhao, et al.
Published: (2025)

Mask Tokens as Prophet: Fine-Grained Cache Eviction for Efficient dLLM Inference
by: Huang, Jianuo, et al.
Published: (2025)

Flash-Unified: A Training-Free and Task-Aware Acceleration Framework for Native Unified Models
by: Ke, Junlong, et al.
Published: (2026)

AgentSlimming: Towards Efficient and Cost-Aware Multi-Agent Systems
by: Chen, Yulang, et al.
Published: (2026)

Efficient Multi-modal Large Language Models via Progressive Consistency Distillation
by: Wen, Zichen, et al.
Published: (2025)

IS-Bench: Evaluating Interactive Safety of VLM-Driven Embodied Agents in Daily Household Tasks
by: Lu, Xiaoya, et al.
Published: (2025)

Prune2Drive: A Plug-and-Play Framework for Accelerating Vision-Language Models in Autonomous Driving
by: Xiong, Minhao, et al.
Published: (2025)

Shifting AI Efficiency From Model-Centric to Data-Centric Compression
by: Liu, Xuyang, et al.
Published: (2025)

HomeGuard: VLM-based Embodied Safeguard for Identifying Contextual Risk in Household Task
by: Lu, Xiaoya, et al.
Published: (2026)

LED-Merging: Mitigating Safety-Utility Conflicts in Model Merging with Location-Election-Disjoint
by: Ma, Qianli, et al.
Published: (2025)

OmniDocLayout: Towards Diverse Document Layout Generation via Coarse-to-Fine LLM Learning
by: Kang, Hengrui, et al.
Published: (2025)

Stop Looking for Important Tokens in Multimodal Language Models: Duplication Matters More
by: Wen, Zichen, et al.
Published: (2025)

AudioMarathon: A Comprehensive Benchmark for Long-Context Audio Understanding and Efficiency in Audio LLMs
by: He, Peize, et al.
Published: (2025)

Diffusion LLM with Native Variable Generation Lengths: Let [EOS] Lead the Way
by: Yang, Yicun, et al.
Published: (2025)

dVLA: Diffusion Vision-Language-Action Model with Multimodal Chain-of-Thought
by: Wen, Junjie, et al.
Published: (2025)

dLLM-Cache: Accelerating Diffusion Large Language Models with Adaptive Caching
by: Liu, Zhiyuan, et al.
Published: (2025)

REEF: Representation Encoding Fingerprints for Large Language Models
by: Zhang, Jie, et al.
Published: (2024)

GraphKV: Breaking the Static Selection Paradigm with Graph-Based KV Cache Eviction
by: Li, Xuelin, et al.
Published: (2025)

Measurement and evaluation of surgical models of spinal fusion in mice, rats, and rabbits
by: Jianghong Lu, et al.
Published: (2026)

Accelerating Diffusion Large Language Models with SlowFast Sampling: The Three Golden Principles
by: Wei, Qingyan, et al.
Published: (2025)

Data Whisperer: Efficient Data Selection for Task-Specific LLM Fine-Tuning via Few-Shot In-Context Learning
by: Wang, Shaobo, et al.
Published: (2025)

Loop as a Bridge: Can Looped Transformers Truly Link Representation Space and Natural Language Outputs?
by: Chen, Guanxu, et al.
Published: (2026)

PPCR-IM: A System for Multi-layer DAG-based Public Policy Consequence Reasoning and Social Indicator Mapping
by: Song, Zichen, et al.
Published: (2026)

Video Compression Commander: Plug-and-Play Inference Acceleration for Video Large Language Models
by: Liu, Xuyang, et al.
Published: (2025)

Mixing Importance with Diversity: Joint Optimization for KV Cache Compression in Large Vision-Language Models
by: Liu, Xuyang, et al.
Published: (2025)

ABSTRAL: Automatic Design of Multi-Agent Systems Through Iterative Refinement and Topology Optimization
by: Song, Weijia, et al.
Published: (2026)

Accelerating Diffusion Transformers with Token-wise Feature Caching
by: Zou, Chang, et al.
Published: (2024)

AudioKV: KV Cache Eviction in Efficient Large Audio Language Models
by: Wang, Yuxuan, et al.
Published: (2026)

Code for the ETAS model used in "Statistical Characteristics of Seismicity Correlated With Crustal Fluids in the Noto Region"
by: Yicun, Guo
Published: (2025)

RiOSWorld: Benchmarking the Risk of Multimodal Computer-Use Agents
by: Yang, Jingyi, et al.
Published: (2025)

EvoStreaming: Your Offline Video Model Is a Natively Streaming Assistant
by: Wen, Zichen, et al.
Published: (2026)