DiffGuard: Text-Based Safety Checker for Diffusion Models

TL;DR

This paper introduces DiffGuard, a novel text-based safety filter for diffusion models that significantly improves the filtering of unsafe content, addressing limitations of current ethical filters in open-source image generation models.

Contribution

The paper presents a new safety filter for diffusion models that outperforms existing solutions by over 14%, enhancing ethical content filtering in AI image generation.

Findings

DiffGuard surpasses existing filters by over 14% in performance.

Current ethical filters have notable limitations in preventing unsafe content.

DiffGuard effectively addresses misuse of AI-generated images in information warfare.

Abstract

Recent advances in Diffusion Models have enabled the generation of images from text, with powerful closed-source models like DALL-E and Midjourney leading the way. However, open-source alternatives, such as StabilityAI's Stable Diffusion, offer comparable capabilities. These open-source models, hosted on Hugging Face, come equipped with ethical filter protections designed to prevent the generation of explicit images. This paper reveals first their limitations and then presents a novel text-based safety filter that outperforms existing solutions. Our research is driven by the critical need to address the misuse of AI-generated content, especially in the context of information warfare. DiffGuard enhances filtering efficacy, achieving a performance that surpasses the best existing filters by over 14%.