VidDoS: Universal Denial-of-Service Attack on Video-based Large Language Models

TL;DR

VidDoS introduces a universal energy-latency attack on Video-LLMs that significantly degrades performance and increases inference latency, posing safety risks in critical applications.

Contribution

This paper presents the first universal ELA framework for Video-LLMs using instance-agnostic triggers and novel optimization techniques, bypassing real-time constraints.

Findings

Induces over 205× token expansion

Increases inference latency by more than 15×

Causes safety violations in autonomous driving simulations

Abstract

Video-LLMs are increasingly deployed in safety-critical applications but are vulnerable to Energy-Latency Attacks (ELAs) that exhaust computational resources. Current image-centric methods fail because temporal aggregation mechanisms dilute individual frame perturbations. Additionally, real-time demands make instance-wise optimization impractical for continuous video streams. We introduce VidDoS, which is the first universal ELA framework tailored for Video-LLMs. Our method leverages universal optimization to create instance-agnostic triggers that require no inference-time gradient calculation. We achieve this through $\textit{masked teacher forcing}$ to steer models toward expensive target sequences, combined with a $\textit{refusal penalty}$ and $\textit{early-termination suppression}$ to override conciseness priors. Testing across three mainstream Video-LLMs and three video datasets, which include video question answering and autonomous driving scenarios, shows extreme degradation. VidDoS induces a token expansion of more than 205$\times$ and inflates the inference latency by more than 15$\times$ relative to clean baselines. Simulations of real-time autonomous driving streams further reveal that this induced latency leads to critical safety violations. We urge the community to recognize and mitigate these high-hazard ELA in Video-LLMs.