Output Scouting: Auditing Large Language Models for Catastrophic Responses

Open in new window