GIFT: Gradient-aware Immunization of diffusion models against malicious Fine-Tuning with safe concepts retention

Open in new window