Energy-Latency Manipulation of Multi-modal Large Language Models via Verbose Samples

TL;DR

This paper explores how to induce high energy and latency costs in multi-modal large language models by crafting verbose samples, thereby revealing a vulnerability that can exhaust computational resources during inference.

Contribution

The authors propose a novel method to manipulate energy-latency costs in MLLMs using verbose samples with specific losses and a balancing algorithm, highlighting a new security concern.

Findings

Verbose samples significantly extend sequence length.

The proposed losses effectively increase energy-latency costs.

The method exposes a vulnerability in MLLMs' deployment security.

Abstract

Despite the exceptional performance of multi-modal large language models (MLLMs), their deployment requires substantial computational resources. Once malicious users induce high energy consumption and latency time (energy-latency cost), it will exhaust computational resources and harm availability of service. In this paper, we investigate this vulnerability for MLLMs, particularly image-based and video-based ones, and aim to induce high energy-latency cost during inference by crafting an imperceptible perturbation. We find that high energy-latency cost can be manipulated by maximizing the length of generated sequences, which motivates us to propose verbose samples, including verbose images and videos. Concretely, two modality non-specific losses are proposed, including a loss to delay end-of-sequence (EOS) token and an uncertainty loss to increase the uncertainty over each generated token. In addition, improving diversity is important to encourage longer responses by increasing the complexity, which inspires the following modality specific loss. For verbose images, a token diversity loss is proposed to promote diverse hidden states. For verbose videos, a frame feature diversity loss is proposed to increase the feature diversity among frames. To balance these losses, we propose a temporal weight adjustment algorithm. Experiments demonstrate that our verbose samples can largely extend the length of generated sequences.