Paper Library

AMV-L: Lifecycle-Managed Agent Memory for Tail-Latency Control in Long-Running LLM Systems

Emmanuel Bamidele

2026-02-22

2603.04443v1

MANATEE: Inference-Time Lightweight Diffusion Based Safety Defense for LLMs

Chun Yan Ryan Kan, Tommy Tran, Vedant Yadav, Ava Cai, Kevin Zhu, Ruizhe Li, Maheep Chaudhary

2026-02-21

safety

2602.18782v1

FENCE: A Financial and Multimodal Jailbreak Detection Dataset

Mirae Kim, Seonghun Jeong, Youngjun Kwak

2026-02-20

red teaming

2602.18154v1

Trojan Horses in Recruiting: A Red-Teaming Case Study on Indirect Prompt Injection in Standard vs. Reasoning Models

Manuel Wirth

2026-02-19

red teaming

2602.18514v1

Fail-Closed Alignment for Large Language Models

Zachary Coalson, Beth Sohler, Aiden Gabriel, Sanghyun Hong

2026-02-19

red teaming

2602.16977v1

Mind the GAP: Text Safety Does Not Transfer to Tool-Call Safety in LLM Agents

Arnold Cartagena, Ariane Teixeira

2026-02-18

safety

2602.16943v1

DeepContext: Stateful Real-Time Detection of Multi-Turn Adversarial Intent Drift in LLMs

Justin Albrethsen, Yash Datta, Kunal Kumar, Sharath Rajasekar

2026-02-18

red teaming

2602.16935v1

NeST: Neuron Selective Tuning for LLM Safety

Sasha Behrouzi, Lichao Wu, Mohamadreza Rostami, Ahmad-Reza Sadeghi

2026-02-18

safety

2602.16835v1

IndicJR: A Judge-Free Benchmark of Jailbreak Robustness in South Asian Languages

Priyaranjan Pattnayak, Sanchari Chowdhuri

2026-02-18

red teaming

2602.16832v1

Policy Compiler for Secure Agentic Systems

Nils Palumbo, Sarthak Choudhary, Jihye Choi, Prasad Chalasani, Somesh Jha

2026-02-18

2602.16708v2

Policy Compiler for Secure Agentic Systems

Nils Palumbo, Sarthak Choudhary, Jihye Choi, Prasad Chalasani, Mihai Christodorescu, Somesh Jha

2026-02-18

2602.16708v1

Align Once, Benefit Multilingually: Enforcing Multilingual Consistency for LLM Safety Alignment

Yuyan Bu, Xiaohao Liu, ZhaoXing Ren, Yaodong Yang, Juntao Dai

2026-02-18

safety

2602.16660v1

Recursive language models for jailbreak detection: a procedural defense for tool-augmented agents

Doron Shavit

2026-02-18

red teaming

2602.16520v1

Helpful to a Fault: Measuring Illicit Assistance in Multi-Turn, Multilingual LLM Agents

Nivya Talokar, Ayush K Tarun, Murari Mandal, Maksym Andriushchenko, Antoine Bosselut

2026-02-18

red teaming

2602.16346v2

February 23 - March 01, 2026

ICON: Indirect Prompt Injection Defense for Agents based on Inference-Time Correction

An LLM-driven Scenario Generation Pipeline Using an Extended Scenic DSL for Autonomous Driving Safety Validation

Skill-Inject: Measuring Agent Vulnerability to Skill File Attacks

Skill-Inject: Measuring Agent Vulnerability to Skill File Attacks

Skill-Inject: Measuring Agent Vulnerability to Skill File Attacks

BarrierSteer: LLM Safety via Learning Barrier Steering

The LLMbda Calculus: AI Agents, Conversations, and Information Flow

Beyond the Binary: A nuanced path for open-weight advanced AI

CIBER: A Comprehensive Benchmark for Security Evaluation of Code Interpreter Agents

Hiding in Plain Text: Detecting Concealed Jailbreaks via Activation Disentanglement

February 16 - February 22, 2026

AMV-L: Lifecycle-Managed Agent Memory for Tail-Latency Control in Long-Running LLM Systems

MANATEE: Inference-Time Lightweight Diffusion Based Safety Defense for LLMs

FENCE: A Financial and Multimodal Jailbreak Detection Dataset

Trojan Horses in Recruiting: A Red-Teaming Case Study on Indirect Prompt Injection in Standard vs. Reasoning Models

Fail-Closed Alignment for Large Language Models

Mind the GAP: Text Safety Does Not Transfer to Tool-Call Safety in LLM Agents

DeepContext: Stateful Real-Time Detection of Multi-Turn Adversarial Intent Drift in LLMs

NeST: Neuron Selective Tuning for LLM Safety

IndicJR: A Judge-Free Benchmark of Jailbreak Robustness in South Asian Languages

Policy Compiler for Secure Agentic Systems

Policy Compiler for Secure Agentic Systems

Align Once, Benefit Multilingually: Enforcing Multilingual Consistency for LLM Safety Alignment

Recursive language models for jailbreak detection: a procedural defense for tool-augmented agents

Helpful to a Fault: Measuring Illicit Assistance in Multi-Turn, Multilingual LLM Agents