Paper Library

Not All Tokens Are Created Equal: Query-Efficient Jailbreak Fuzzing for LLMs

Wenyu Chen, Xiangtao Meng, Chuanchao Zang, Li Wang, Xinyu Gao, Jianing Wang, Peng Zhan, Zheng Li, Shanqing Guo

2026-03-24

red teaming

2603.23269v1

SafeSeek: Universal Attribution of Safety Circuits in Language Models

Miao Yu, Siyuan Fu, Moayad Aloqaily, Zhenhong Zhou, Safa Otoum, Xing fan, Kun Wang, Yufei Guo, Qingsong Wen

2026-03-24

red teaming

2603.23268v1

Mind Your HEARTBEAT! Claw Background Execution Inherently Enables Silent Memory Pollution

Yechao Zhang, Shiqian Zhao, Jie Zhang, Gelei Deng, Jiawen Zhang, Xiaogeng Liu, Chaowei Xiao, Tianwei Zhang

2026-03-24

2603.23064v3

Mind Your HEARTBEAT! Claw Background Execution Inherently Enables Silent Memory Pollution

Yechao Zhang, Shiqian Zhao, Jie Zhang, Gelei Deng, Jiawen Zhang, Xiaogeng Liu, Chaowei Xiao, Tianwei Zhang

2026-03-24

2603.23064v2

Mind Your HEARTBEAT! Claw Background Execution Inherently Enables Silent Memory Pollution

Yechao Zhang, Shiqian Zhao, Jie Zhang, Gelei Deng, Jiawen Zhang, Xiaogeng Liu, Chaowei Xiao, Tianwei Zhang

2026-03-24

2603.23064v1

SoK: The Attack Surface of Agentic AI -- Tools, and Autonomy

Ali Dehghantanha, Sajad Homayoun

2026-03-24

red teaming

2603.22928v1

TreeTeaming: Autonomous Red-Teaming of Vision-Language Models via Hierarchical Strategy Exploration

Chunxiao Li, Lijun Li, Jing Shao

2026-03-24

red teaming

2603.22882v1

LLMON: An LLM-native Markup Language to Leverage Structure and Semantics at the LLM Interface

Michael Hind, Basel Shbita, Bo Wu, Farhan Ahmed, Chad DeLuca, Nathan Fulton, David Cox, Dan Gutfreund

2026-03-23

2603.22519v2

LLMON: An LLM-native Markup Language to Leverage Structure and Semantics at the LLM Interface

Michael Hind, Basel Shbita, Bo Wu, Farhan Ahmed, Chad DeLuca, Nathan Fulton, David Cox, Dan Gutfreund

2026-03-23

2603.22519v1

Model Context Protocol Threat Modeling and Analyzing Vulnerabilities to Prompt Injection with Tool Poisoning

Charoes Huang, Xin Huang, Ngoc Phu Tran, Amin Milani Fard

2026-03-23

red teaming

2603.22489v1

Principled Steering via Null-space Projection for Jailbreak Defense in Vision-Language Models

Xingyu Zhu, Beier Zhu, Shuo Wang, Junfeng Fang, Kesen Zhao, Hanwang Zhang, Xiangnan He

2026-03-23

red teaming

2603.22094v2

Principled Steering via Null-space Projection for Jailbreak Defense in Vision-Language Models

Xingyu Zhu, Beier Zhu, Shuo Wang, Junfeng Fang, Kesen Zhao, Hanwang Zhang, Xiangnan He

2026-03-23

2603.22094v1

SecureBreak -- A dataset towards safe and secure models

Marco Arazzi, Vignesh Kumar Kembu, Antonino Nocera

2026-03-23

red teaming

2603.21975v1

Structured Visual Narratives Undermine Safety Alignment in Multimodal Large Language Models

Rui Yang Tan, Yujia Hu, Roy Ka-Wei Lee

2026-03-23

red teaming

2603.21697v1

Are AI-assisted Development Tools Immune to Prompt Injection?

Charoes Huang, Xin Huang, Amin Milani Fard

2026-03-23

red teaming

2603.21642v1

March 23 - March 29, 2026

Not All Tokens Are Created Equal: Query-Efficient Jailbreak Fuzzing for LLMs

SafeSeek: Universal Attribution of Safety Circuits in Language Models

Mind Your HEARTBEAT! Claw Background Execution Inherently Enables Silent Memory Pollution

Mind Your HEARTBEAT! Claw Background Execution Inherently Enables Silent Memory Pollution

Mind Your HEARTBEAT! Claw Background Execution Inherently Enables Silent Memory Pollution

SoK: The Attack Surface of Agentic AI -- Tools, and Autonomy

TreeTeaming: Autonomous Red-Teaming of Vision-Language Models via Hierarchical Strategy Exploration

LLMON: An LLM-native Markup Language to Leverage Structure and Semantics at the LLM Interface

LLMON: An LLM-native Markup Language to Leverage Structure and Semantics at the LLM Interface

Model Context Protocol Threat Modeling and Analyzing Vulnerabilities to Prompt Injection with Tool Poisoning

Principled Steering via Null-space Projection for Jailbreak Defense in Vision-Language Models

Principled Steering via Null-space Projection for Jailbreak Defense in Vision-Language Models

SecureBreak -- A dataset towards safe and secure models

Structured Visual Narratives Undermine Safety Alignment in Multimodal Large Language Models

Are AI-assisted Development Tools Immune to Prompt Injection?

March 16 - March 22, 2026

The Workload-Router-Pool Architecture for LLM Inference Optimization: A Vision Paper from the vLLM Semantic Router Project

JANUS: A Lightweight Framework for Jailbreaking Text-to-Image Models via Distribution Optimization

JANUS: A Lightweight Framework for Jailbreaking Text-to-Image Models via Distribution Optimization

Detection of adversarial intent in Human-AI teams using LLMs

The production of meaning in the processing of natural language

Evolving Jailbreaks: Automated Multi-Objective Long-Tail Attacks on Large Language Models

Trojan's Whisper: Stealthy Manipulation of OpenClaw through Injected Bootstrapped Guidance

The Causal Impact of Tool Affordance on Safety Alignment in LLM Agents

A Framework for Formalizing LLM Agent Security