Phi: Preference Hijacking in Multi-modal Large Language Models at Inference Time

ACL ARR 2025 May Submission3621 Authors

Abstract:

Recently, Multimodal Large Language Models (MLLMs) have gained significant attention across various domains. However, their widespread adoption has also raised serious safety concerns.
In this paper, we uncover a new safety risk of MLLMs: the output preference of MLLMs can be arbitrarily manipulated by carefully optimized images. Such attacks often generate contextually relevant yet biased responses that are neither overtly harmful nor unethical, making them difficult to detect. Specifically, we introduce a novel method, **P**reference **Hi**jacking (**Phi**), for manipulating the MLLM response preferences using a preference hijacked image. Our method works at inference time and requires no model modifications. Additionally, we introduce a universal hijacking perturbation -- a transferable component that can be embedded into different images to hijack MLLM responses toward any attacker-specified preferences. Experimental results across various tasks demonstrate the effectiveness of our approach.

Paper Type: Long

Research Area: Interpretability and Analysis of Models for NLP

Research Area Keywords: Multi-modal Large Language Models, Model Preference, Model Safety

Contribution Types: Model analysis & interpretability

Languages Studied: English

Submission Number: 3621