Your question is Evaluation Metrics for GenAI Security. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
Explain the evaluation metrics you would use to validate the precision, recall, and robustness of a generative AI security tool.
Discuss how you would measure correct detections, missed threats, false alarms, and resilience to adversarial or changing inputs. Describe the evaluation dataset, segmentation strategy, human review process, and acceptance criteria you would use. Explain how you would handle nondeterministic model outputs and balance security coverage against usability.