Prompt-Based Safety Guidance Is Ineffective for Unlearned Text-to-Image Diffusion Models

Open in new window