Towards Understanding Unsafe Video Generation

TL;DR

This paper investigates the potential for video generation models to produce unsafe content, creating a new dataset of unsafe videos and proposing a novel defense mechanism called Latent Variable Defense to mitigate such risks.

Contribution

It provides the first dataset of unsafe videos generated by VGMs and introduces Latent Variable Defense, a new method to prevent unsafe video generation within the model's sampling process.

Findings

Generated 2112 unsafe videos from open-source models and prompts.

Identified five unsafe video categories through analysis.

Proposed LVD achieves 0.90 defense accuracy with 10x efficiency.

Abstract

Video generation models (VGMs) have demonstrated the capability to synthesize high-quality output. It is important to understand their potential to produce unsafe content, such as violent or terrifying videos. In this work, we provide a comprehensive understanding of unsafe video generation. First, to confirm the possibility that these models could indeed generate unsafe videos, we choose unsafe content generation prompts collected from 4chan and Lexica, and three open-source SOTA VGMs to generate unsafe videos. After filtering out duplicates and poorly generated content, we created an initial set of 2112 unsafe videos from an original pool of 5607 videos. Through clustering and thematic coding analysis of these generated videos, we identify 5 unsafe video categories: Distorted/Weird, Terrifying, Pornographic, Violent/Bloody, and Political. With IRB approval, we then recruit online participants to help label the generated videos. Based on the annotations submitted by 403 participants, we identified 937 unsafe videos from the initial video set. With the labeled information and the corresponding prompts, we created the first dataset of unsafe videos generated by VGMs. We then study possible defense mechanisms to prevent the generation of unsafe videos. Existing defense methods in image generation focus on filtering either input prompt or output results. We propose a new approach called Latent Variable Defense (LVD), which works within the model's internal sampling process. LVD can achieve 0.90 defense accuracy while reducing time and computing resources by 10x when sampling a large number of unsafe prompts.