English

Mitigating Backdoor Threats to Large Language Models: Advancement and Challenges

Cryptography and Security 2024-10-01 v1 Artificial Intelligence Computation and Language Machine Learning Systems and Control Systems and Control

Abstract

The advancement of Large Language Models (LLMs) has significantly impacted various domains, including Web search, healthcare, and software development. However, as these models scale, they become more vulnerable to cybersecurity risks, particularly backdoor attacks. By exploiting the potent memorization capacity of LLMs, adversaries can easily inject backdoors into LLMs by manipulating a small portion of training data, leading to malicious behaviors in downstream applications whenever the hidden backdoor is activated by the pre-defined triggers. Moreover, emerging learning paradigms like instruction tuning and reinforcement learning from human feedback (RLHF) exacerbate these risks as they rely heavily on crowdsourced data and human feedback, which are not fully controlled. In this paper, we present a comprehensive survey of emerging backdoor threats to LLMs that appear during LLM development or inference, and cover recent advancement in both defense and detection strategies for mitigating backdoor threats to LLMs. We also outline key challenges in addressing these threats, highlighting areas for future research.

Keywords

Cite

@article{arxiv.2409.19993,
  title  = {Mitigating Backdoor Threats to Large Language Models: Advancement and Challenges},
  author = {Qin Liu and Wenjie Mo and Terry Tong and Jiashu Xu and Fei Wang and Chaowei Xiao and Muhao Chen},
  journal= {arXiv preprint arXiv:2409.19993},
  year   = {2024}
}

Comments

The 60th Annual Allerton Conference (Invited Paper). The arXiv version is a pre-IEEE Press publication version

R2 v1 2026-06-28T19:01:46.112Z