A company implements a security system that tracks how many requests users send to an AI model per minute. If a user exceeds the threshold, further requests are temporarily blocked. Which monitoring control does this represent?

A cloud-based AI service enforces user authentication tokens before allowing model access. Each endpoint has separate permissions defining which users can submit prompts or retrieve results. What mechanism does this describe?

A healthcare chatbot flags low-confidence answers for manual verification and limits users to five queries per minute to prevent overload. Which two prompt monitoring mechanisms are applied?

An e-commerce platform's AI assistant has been manipulated by customers who discovered that certain user prompt patterns can extract the system's internal pricing algorithms and discount rules. Investigation shows the system prompt contained these rules for context, making them vulnerable to extraction. What fundamental system prompt security principle was violated?

A financial AI chatbot is tricked by a user into revealing confidential internal policies by inserting a hidden instruction within a normal-looking question. Which type of attack has occurred?