Your company requires that no "Human" at Microsoft can ever see the prompts sent to the AI. Which feature should you request/enable?
check_circle
Correct AnswerD
Certain regulated industries can apply for an "Abuse Monitoring" exemption to ensure prompts/outputs are not stored for human review.
Question 282
A bank wants to use AI to detect fraudulent transactions. Is a Large Language Model (LLM) the best primary tool for this?
check_circle
Correct AnswerA
LLMs are for unstructured text; traditional machine learning is much better at identifying patterns in structured transaction data.
Question 283
Your "Run" status is failed with a content_filter message. Is this a system bug?
check_circle
Correct AnswerC
This is a safety feature, not a bug; it indicates the content was blocked by Azure AI Content Safety.
Question 284
You are pitching a RAG solution to a legal department. What is the primary business value compared to a standard LLM?
check_circle
Correct AnswerD
Grounding the AI in specific, verified documents is the most critical requirement for high-stakes industries like Law.
Question 285
A hospital wants an AI to summarize patient records. Which security standard is non-negotiable for this deployment in the US?
check_circle
Correct AnswerA
HIPAA compliance is required for any system handling protected health information (PHI) in the United States.
Question 286
Why might an agent "Forget" instructions given 20 messages ago in a long thread?
check_circle
Correct AnswerA
As a thread grows, older messages are dropped or truncated to fit within the model's limited token capacity (context).
Question 287
Which feature in Document Intelligence allows you to detect "Checkboxes" and "Radio Buttons" (Selection Marks) on a filled-out survey?
check_circle
Correct AnswerA
The Layout model extracts the geometric structure of a document, including tables, selection marks, and text lines.
Question 288
What is "Data at Rest" encryption in Azure AI Search?
check_circle
Correct AnswerD
All search indexes are encrypted by default, but Customer-Managed Keys (CMK) provide an extra layer of control.
Question 289
You are building an agent for a global NGO. They need it to work in languages with very little data (Low-resource languages). What is the best strategy?
check_circle
Correct AnswerC
Translating the user's input to a high-resource language like English before processing often yields better results for rare languages.
Question 290
You want to evaluate if an agent's response effectively addresses the user's intent. Which AI Foundry metric should you use?
check_circle
Correct AnswerC
Relevance specifically compares the model's response to the user's prompt to ensure it is helpful and on-topic.
