Multimodal Pragmatic Jailbreak on Text-to-image Models

Open in new window