Your question is Model Observability and Drift. 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).
How do you ensure model observability and drift detection once an AI-driven security product is in production?