Training-Free Safe Denoisers for Safe Use of Diffusion Models

TL;DR

This paper introduces a training-free method to modify diffusion model sampling trajectories, ensuring generated images avoid unsafe or unwanted data regions without retraining or fine-tuning.

Contribution

It proposes a novel approach that directly adjusts the sampling process using a negation set, eliminating the need for retraining DMs to enhance safety.

Findings

Successfully generates high-quality safe images avoiding negation areas

Applicable to text-conditional, class-conditional, and unconditional generation

Demonstrates potential for safer diffusion model applications

Abstract

There is growing concern over the safety of powerful diffusion models (DMs), as they are often misused to produce inappropriate, not-safe-for-work (NSFW) content or generate copyrighted material or data of individuals who wish to be forgotten. Many existing methods tackle these issues by heavily relying on text-based negative prompts or extensively retraining DMs to eliminate certain features or samples. In this paper, we take a radically different approach, directly modifying the sampling trajectory by leveraging a negation set (e.g., unsafe images, copyrighted data, or datapoints needed to be excluded) to avoid specific regions of data distribution, without needing to retrain or fine-tune DMs. We formally derive the relationship between the expected denoised samples that are safe and those that are not safe, leading to our $\textit{safe}$ denoiser which ensures its final samples are away from the area to be negated. Inspired by the derivation, we develop a practical algorithm that successfully produces high-quality samples while avoiding negation areas of the data distribution in text-conditional, class-conditional, and unconditional image generation scenarios. These results hint at the great potential of our training-free safe denoiser for using DMs more safely.