SP-Guard: Selective Prompt-adaptive Guidance for Safe Text-to-Image Generation

TL;DR

SP-Guard introduces a novel method for safer text-to-image generation by adaptively guiding only unsafe regions based on prompt harmfulness, enhancing safety and controllability in diffusion models.

Contribution

It proposes a prompt-aware, selective guidance mechanism that improves safety without compromising image quality in diffusion-based T2I models.

Findings

SP-Guard generates safer images than existing methods.

It minimizes unintended alterations during safe image generation.

The approach emphasizes transparency and controllability in AI-generated images.

Abstract

While diffusion-based T2I models have achieved remarkable image generation quality, they also enable easy creation of harmful content, raising social concerns and highlighting the need for safer generation. Existing inference-time guiding methods lack both adaptivity--adjusting guidance strength based on the prompt--and selectivity--targeting only unsafe regions of the image. Our method, SP-Guard, addresses these limitations by estimating prompt harmfulness and applying a selective guidance mask to guide only unsafe areas. Experiments show that SP-Guard generates safer images than existing methods while minimizing unintended content alteration. Beyond improving safety, our findings highlight the importance of transparency and controllability in image generation.