Cover Image

The Autonomy Tax: Defense Training Breaks LLM Agents

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Li, Shawn, Zhao, Yue
Format:	Preprint
Published:	2026
Subjects:	Cryptography and Security Artificial Intelligence Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Training RL Agents for Multi-Objective Network Defense Tasks
by: Molina-Markham, Andres, et al.
Published: (2025)

A Survey on Model Extraction Attacks and Defenses for Large Language Models
by: Zhao, Kaixiang, et al.
Published: (2025)

A Survey of Model Extraction Attacks and Defenses in Distributed Computing Environments
by: Zhao, Kaixiang, et al.
Published: (2025)

A Systematic Survey of Model Extraction Attacks and Defenses: State-of-the-Art and Perspectives
by: Zhao, Kaixiang, et al.
Published: (2025)

Attacks and Defenses Against LLM Fingerprinting
by: Kurian, Kevin, et al.
Published: (2025)

TimeGuard: Channel-wise Pool Training for Backdoor Defense in Time Series Forecasting
by: Nguyen, Quang Duc, et al.
Published: (2026)

Breaking Agent Backbones: Evaluating the Security of Backbone LLMs in AI Agents
by: Bazinska, Julia, et al.
Published: (2025)

Seed Hijacking of LLM Sampling and Quantum Random Number Defense
by: You, Ziyang, et al.
Published: (2026)

RICE: Breaking Through the Training Bottlenecks of Reinforcement Learning with Explanation
by: Cheng, Zelei, et al.
Published: (2024)

PoolFlip: A Multi-Agent Reinforcement Learning Security Environment for Cyber Defense
by: Cadet, Xavier, et al.
Published: (2025)

LeakAgent: RL-based Red-teaming Agent for LLM Privacy Leakage
by: Nie, Yuzhou, et al.
Published: (2024)

Evaluating Prompt Injection Defenses for Educational LLM Tutors: Security-Usability-Latency Trade-offs
by: Maiorano, Alexandre Cristovão
Published: (2026)

The Jailbreak Tax: How Useful are Your Jailbreak Outputs?
by: Nikolić, Kristina, et al.
Published: (2025)

Adversarial Reinforcement Learning for Offensive and Defensive Agents in a Simulated Zero-Sum Network Environment
by: Shahid, Abrar, et al.
Published: (2025)

PureGen: Universal Data Purification for Train-Time Poison Defense via Generative Model Dynamics
by: Bhat, Sunay, et al.
Published: (2024)

Defense effectiveness across architectural layers: a mechanistic evaluation of persistent memory attacks on stateful LLM agents
by: Leong, Jun Wen
Published: (2026)

Uncovering, Explaining, and Mitigating the Superficial Safety of Backdoor Defense
by: Min, Rui, et al.
Published: (2024)

Reliable Weak-to-Strong Monitoring of LLM Agents
by: Kale, Neil, et al.
Published: (2025)

Safety Tax: Safety Alignment Makes Your Large Reasoning Models Less Reasonable
by: Huang, Tiansheng, et al.
Published: (2025)

A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment
by: Wang, Kun, et al.
Published: (2025)

Efficient but Vulnerable: Benchmarking and Defending LLM Batch Prompting Attack
by: Yue, Murong, et al.
Published: (2025)

Optimal Defenses Against Gradient Reconstruction Attacks
by: Chen, Yuxiao, et al.
Published: (2024)

Mitigating the Structural Bias in Graph Adversarial Defenses
by: Fang, Junyuan, et al.
Published: (2025)

SHADE-Arena: Evaluating Sabotage and Monitoring in LLM Agents
by: Kutasov, Jonathan, et al.
Published: (2025)

OEP: Poisoning Self-Evolving LLM Agents via Locally Correct but Non-Transferable Experiences
by: Wang, Kaixiang, et al.
Published: (2026)

UDora: A Unified Red Teaming Framework against LLM Agents by Dynamically Hijacking Their Own Reasoning
by: Zhang, Jiawei, et al.
Published: (2025)

Defenses Against Prompt Attacks Learn Surface Heuristics
by: Li, Shawn, et al.
Published: (2026)

A Causal Perspective for Enhancing Jailbreak Attack and Defense
by: Pan, Licheng, et al.
Published: (2026)

Stealthy Poisoning Attacks Bypass Defenses in Regression Settings
by: Carnerero-Cano, Javier, et al.
Published: (2026)

HARP: Measuring Harm Amplification in Multi-Agent LLM Systems
by: Rahman, Md Hafizur, et al.
Published: (2026)

SHIELD: Secure Hypernetworks for Incremental Expansion Learning Defense
by: Krukowski, Patryk, et al.
Published: (2025)

Security in the Fine-Tuning Lifecycle of Large Language Models: Threats, Defenses,Evaluation, and Future Directions
by: Li, Wenjuan, et al.
Published: (2026)

Can Adversarial Code Comments Fool AI Security Reviewers -- Large-Scale Empirical Study of Comment-Based Attacks and Defenses Against LLM Code Analysis
by: Thornton, Scott
Published: (2026)

Mjolnir: Breaking the Shield of Perturbation-Protected Gradients via Adaptive Diffusion
by: Liu, Xuan, et al.
Published: (2024)

Large Language Models in Cybersecurity: Applications, Vulnerabilities, and Defense Techniques
by: Jaffal, Niveen O., et al.
Published: (2025)

SoK: Benchmarking Poisoning Attacks and Defenses in Federated Learning
by: Zhang, Heyi, et al.
Published: (2025)

Attacks and Defenses for Generative Diffusion Models: A Comprehensive Survey
by: Truong, Vu Tuan, et al.
Published: (2024)

Combining Stochastic Defenses to Resist Gradient Inversion: An Ablation Study
by: Scheliga, Daniel, et al.
Published: (2022)

Fuzz-Testing Meets LLM-Based Agents: An Automated and Efficient Framework for Jailbreaking Text-To-Image Generation Models
by: Dong, Yingkai, et al.
Published: (2024)

Designing Robust Cyber-Defense Agents with Evolving Behavior Trees
by: Potteiger, Nicholas, et al.
Published: (2024)