Merge Hijacking: Backdoor Attacks to Model Merging of Large Language Models
Preprint
Abstract
Model merging for Large Language Models (LLMs) directly fuses the parameters of different models finetuned on various tasks, creating a unified model for multi-domain tasks. However, due to potential vulnerabilities in models available on open-source platforms, model merging is susceptible to backdoor attacks. In this paper, we propose Merge Hijacking, the first backdoor attack targeting model merging in LLMs. The attacker constructs a malicious upload model and releases it. Once a victim user merges it with any other models, the resulting merged model inherits the backdoor while maintaining utility across tasks. Merge Hijacking defines two main objectives-effectiveness and utility-and achieves them through four steps. Extensive experiments demonstrate the effectiveness of our attack across different models, merging algorithms, and tasks. Additionally, we show that the attack remains effective even when merging real-world models. Moreover, our attack demonstrates robustness against two inference-time defenses (Paraphrasing and CLEANGEN) and one training-time defense (Fine-pruning).Keywords
All Related Versions
This publication has 8 references indexed in Scilit:
- Model Merging in LLMs, MLLMs, and Beyond: Methods, Theories, Applications, and OpportunitiesACM Computing Surveys, 2026
- BadToken: Token-level Backdoor Attacks to Multi-modal Large Language Models2025
- CleanGen: Mitigating Backdoor Attacks for Generation Tasks in Large Language Models2024
- Here’s a Free Lunch: Sanitizing Backdoored Models with Model Merge2024
- BadAgent: Inserting and Activating Backdoor Attacks in LLM Agents2024
- Composite Backdoor Attacks Against Large Language Models2024
- Backdoor Pre-trained Models Can Transfer to All2021
- Hidden Killer: Invisible Textual Backdoor Attacks with Syntactic Trigger2021