On the Proactive Generation of Unsafe Images From Text-To-Image Models Using Benign Prompts

Last updated