Confidence Elicitation: A New Attack Vector for Large Language Models

Open in new window