Cover Image

From Poisoned to Aware: Fostering Backdoor Self-Awareness in LLMs

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Shen, Guangyu, Cheng, Siyuan, Xu, Xiangzhe, Zhou, Yuan, Guo, Hanxi, Zhang, Zhuo, Zhang, Xiangyu
Format:	Preprint
Published:	2025
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

MGC: A Compiler Framework Exploiting Compositional Blindness in Aligned LLMs for Malware Generation
by: Yan, Lu, et al.
Published: (2025)

Backdoors in RLVR: Jailbreak Backdoors in LLMs From Verifiable Reward
by: Guo, Weiyang, et al.
Published: (2026)

ASPIRER: Bypassing System Prompts With Permutation-based Backdoors in LLMs
by: Yan, Lu, et al.
Published: (2024)

Rapid Optimization for Jailbreaking LLMs via Subconscious Exploitation and Echopraxia
by: Shen, Guangyu, et al.
Published: (2024)

MARS: A Malignity-Aware Backdoor Defense in Federated Learning
by: Wan, Wei, et al.
Published: (2025)

Elijah: Eliminating Backdoors Injected in Diffusion Models via Distribution Shift
by: An, Shengwei, et al.
Published: (2023)

P2P: A Poison-to-Poison Remedy for Reliable Backdoor Defense in LLMs
by: Zhao, Shuai, et al.
Published: (2025)

UNIT: Backdoor Mitigation via Automated Neural Distribution Tightening
by: Cheng, Siyuan, et al.
Published: (2024)

SOFT: Selective Data Obfuscation for Protecting LLM Fine-tuning against Membership Inference Attacks
by: Zhang, Kaiyuan, et al.
Published: (2025)

The Stronger the Diffusion Model, the Easier the Backdoor: Data Poisoning to Induce Copyright Breaches Without Adjusting Finetuning Pipeline
by: Wang, Haonan, et al.
Published: (2024)

Poison Once, Control Anywhere: Clean-Text Visual Backdoors in VLM-based Mobile Agents
by: Wang, Xuan, et al.
Published: (2025)

BadSkill: Backdoor Attacks on Agent Skills via Model-in-Skill Poisoning
by: Tie, Guiyao, et al.
Published: (2026)

The Ripple Effect: On Unforeseen Complications of Backdoor Attacks
by: Zhang, Rui, et al.
Published: (2025)

Toward Polymorphic Backdoor against Semantic Communication via Intensity-Based Poisoning
by: Yang, Xiao, et al.
Published: (2026)

CBPF: Filtering Poisoned Data Based on Composite Backdoor Attack
by: Xia, Hanfeng, et al.
Published: (2024)

Acquiring Clean Language Models from Backdoor Poisoned Datasets by Downscaling Frequency Space
by: Wu, Zongru, et al.
Published: (2024)

Persistent Pre-Training Poisoning of LLMs
by: Zhang, Yiming, et al.
Published: (2024)

Structure-Aware Distributed Backdoor Attacks in Federated Learning
by: Jian, Wang, et al.
Published: (2026)

Through the Stealth Lens: Attention-Aware Defenses Against Poisoning in RAG
by: Choudhary, Sarthak, et al.
Published: (2025)

LOTUS: Evasive and Resilient Backdoor Attacks through Sub-Partitioning
by: Cheng, Siyuan, et al.
Published: (2024)

Backdoor4Good: Benchmarking Beneficial Uses of Backdoors in LLMs
by: Li, Yige, et al.
Published: (2026)

Does Few-shot Learning Suffer from Backdoor Attacks?
by: Liu, Xinwei, et al.
Published: (2023)

Persistent Backdoor Attacks under Continual Fine-Tuning of LLMs
by: Cui, Jing, et al.
Published: (2025)

On The Dangers of Poisoned LLMs In Security Automation
by: Karlsen, Patrick, et al.
Published: (2025)

Concealing Backdoor Model Updates in Federated Learning by Trigger-Optimized Data Poisoning
by: Zhang, Yujie, et al.
Published: (2024)

Class-Conditional Neural Polarizer: A Lightweight and Effective Backdoor Defense by Purifying Poisoned Features
by: Zhu, Mingli, et al.
Published: (2025)

A Set of Generalized Components to Achieve Effective Poison-only Clean-label Backdoor Attacks with Collaborative Sample Selection and Triggers
by: Wu, Zhixiao, et al.
Published: (2025)

Compromising Embodied Agents with Contextual Backdoor Attacks
by: Liu, Aishan, et al.
Published: (2024)

ME: Trigger Element Combination Backdoor Attack on Copyright Infringement
by: Yang, Feiyu, et al.
Published: (2025)

TASER: Task-Aware Spectral Energy Refine for Backdoor Suppression in UAV Swarms Decentralized Federated Learning
by: Huang, Sizhe, et al.
Published: (2026)

Opening A Pandora's Box: Things You Should Know in the Era of Custom GPTs
by: Tao, Guanhong, et al.
Published: (2023)

Concept-Guided Backdoor Attack on Vision Language Models
by: Shen, Haoyu, et al.
Published: (2025)

CSC: Turning the Adversary's Poison against Itself
by: Shi, Yuchen, et al.
Published: (2026)

A Survey on Backdoor Threats in Large Language Models (LLMs): Attacks, Defenses, and Evaluations
by: Zhou, Yihe, et al.
Published: (2025)

Injecting Universal Jailbreak Backdoors into LLMs in Minutes
by: Chen, Zhuowei, et al.
Published: (2025)

Backdoor Sentinel: Detecting and Detoxifying Backdoors in Diffusion Models via Temporal Noise Consistency
by: Wang, Bingzheng, et al.
Published: (2026)

SAB:A Stealing and Robust Backdoor Attack based on Steganographic Algorithm against Federated Learning
by: Xu, Weida, et al.
Published: (2024)

ASTRA: Autonomous Spatial-Temporal Red-teaming for AI Software Assistants
by: Xu, Xiangzhe, et al.
Published: (2025)

Swallowing the Poison Pills: Insights from Vulnerability Disparity Among LLMs
by: Yifeng, Peng, et al.
Published: (2025)

Defusing the Trigger: Plug-and-Play Defense for Backdoored LLMs via Tail-Risk Intrinsic Geometric Smoothing
by: Fan, Kaisheng, et al.
Published: (2026)