AutoBackdoor: Automating Backdoor Attacks via LLM Agents

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Yige, Li, Zhe, Zhao, Wei, Min, Nay Myat, Huang, Hanxun, Ma, Xingjun, Sun, Jun
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908668128133120
author Li, Yige
Li, Zhe
Zhao, Wei
Min, Nay Myat
Huang, Hanxun
Ma, Xingjun
Sun, Jun
author_facet Li, Yige
Li, Zhe
Zhao, Wei
Min, Nay Myat
Huang, Hanxun
Ma, Xingjun
Sun, Jun
contents Backdoor attacks pose a serious threat to the secure deployment of large language models (LLMs), enabling adversaries to implant hidden behaviors triggered by specific inputs. However, existing methods often rely on manually crafted triggers and static data pipelines, which are rigid, labor-intensive, and inadequate for systematically evaluating modern defense robustness. As AI agents become increasingly capable, there is a growing need for more rigorous, diverse, and scalable \textit{red-teaming frameworks} that can realistically simulate backdoor threats and assess model resilience under adversarial conditions. In this work, we introduce \textsc{AutoBackdoor}, a general framework for automating backdoor injection, encompassing trigger generation, poisoned data construction, and model fine-tuning via an autonomous agent-driven pipeline. Unlike prior approaches, AutoBackdoor uses a powerful language model agent to generate semantically coherent, context-aware trigger phrases, enabling scalable poisoning across arbitrary topics with minimal human effort. We evaluate AutoBackdoor under three realistic threat scenarios, including \textit{Bias Recommendation}, \textit{Hallucination Injection}, and \textit{Peer Review Manipulation}, to simulate a broad range of attacks. Experiments on both open-source and commercial models, including LLaMA-3, Mistral, Qwen, and GPT-4o, demonstrate that our method achieves over 90\% attack success with only a small number of poisoned samples. More importantly, we find that existing defenses often fail to mitigate these attacks, underscoring the need for more rigorous and adaptive evaluation techniques against agent-driven threats as explored in this work. All code, datasets, and experimental configurations will be merged into our primary repository at https://github.com/bboylyg/BackdoorLLM.
format Preprint
id arxiv_https___arxiv_org_abs_2511_16709
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AutoBackdoor: Automating Backdoor Attacks via LLM Agents
Li, Yige
Li, Zhe
Zhao, Wei
Min, Nay Myat
Huang, Hanxun
Ma, Xingjun
Sun, Jun
Cryptography and Security
Artificial Intelligence
Backdoor attacks pose a serious threat to the secure deployment of large language models (LLMs), enabling adversaries to implant hidden behaviors triggered by specific inputs. However, existing methods often rely on manually crafted triggers and static data pipelines, which are rigid, labor-intensive, and inadequate for systematically evaluating modern defense robustness. As AI agents become increasingly capable, there is a growing need for more rigorous, diverse, and scalable \textit{red-teaming frameworks} that can realistically simulate backdoor threats and assess model resilience under adversarial conditions. In this work, we introduce \textsc{AutoBackdoor}, a general framework for automating backdoor injection, encompassing trigger generation, poisoned data construction, and model fine-tuning via an autonomous agent-driven pipeline. Unlike prior approaches, AutoBackdoor uses a powerful language model agent to generate semantically coherent, context-aware trigger phrases, enabling scalable poisoning across arbitrary topics with minimal human effort. We evaluate AutoBackdoor under three realistic threat scenarios, including \textit{Bias Recommendation}, \textit{Hallucination Injection}, and \textit{Peer Review Manipulation}, to simulate a broad range of attacks. Experiments on both open-source and commercial models, including LLaMA-3, Mistral, Qwen, and GPT-4o, demonstrate that our method achieves over 90\% attack success with only a small number of poisoned samples. More importantly, we find that existing defenses often fail to mitigate these attacks, underscoring the need for more rigorous and adaptive evaluation techniques against agent-driven threats as explored in this work. All code, datasets, and experimental configurations will be merged into our primary repository at https://github.com/bboylyg/BackdoorLLM.
title AutoBackdoor: Automating Backdoor Attacks via LLM Agents
topic Cryptography and Security
Artificial Intelligence
url https://arxiv.org/abs/2511.16709