Enhancing Model Defense Against Jailbreaks with Proactive Safety Reasoning

Open in new window