Microsoft AB-730 Online Practice Questions and Exam Preparation

Microsoft AB-730 Online Questions & Answers

• Question 71:

  You are a merchandiser who is planning for the upcoming season.

  You prompt Microsoft 365 Copilot to suggest which products to stock based on historical sales data.

  Without reviewing the suggestions or checking current market trends, you place a large order based solely on the output of Copilot.

  What is this an example of?

  A. verification
  B. overreliance
  C. fabrication
  D. prompt injection
  B. overreliance

  ---

  Explanation

  Overreliance occurs when a user accepts AI output without appropriate human review, validation, or business judgment. In this scenario, the merchandiser places a large order solely because Copilot suggested it and does not check the suggestions against current market trends or other evidence. Fabrication would describe Copilot inventing unsupported information, and prompt injection would involve malicious or manipulative instructions affecting the model. The responsible AI lesson is that Copilot output should inform decisions, not replace review for consequential business actions.

• Question 72:

  HOTSPOT

  You are evaluating the features of Microsoft 365 Copilot notebooks.

  For each of the following statements, select Yes if the statement is true. Otherwise, select No.

  NOTE: Each correct selection is worth one point.

  

  

  ---

  All three statements are selected as Yes. Copilot Notebooks can be used as a basis for creating an agent, can support generated audio overviews, and can continue adding or updating references as the notebook evolves. A notebook is not only a static container; it is a workspace for collecting and using related sources. The exam takeaway is that notebooks support ongoing, source-centered Copilot work.

• Question 73:

  You join an internal Microsoft Teams meeting late and want to catch up on what you missed.

  The meeting is being recorded. You need to summarize the portion of the meeting that you missed as soon as possible.

  What is the best approach to achieve the goal?

  More than one answer choice may achieve the goal. Select the BEST answer.

  A. From the Teams app, ask Copilot in Teams to summarize what you missed so far.
  B. From the meeting, read the transcript.
  C. From the Microsoft 365 Copilot app, ask Copilot to tell you what you missed.
  D. From the Teams app, view the intelligent recap.
  D. From the Teams app, view the intelligent recap.

  ---

  Explanation

  Modify answer: A

  Because the user joined the meeting late and needs the missed portion as soon as possible, the best action is to ask Copilot in Teams during the meeting to summarize what was missed so far. Intelligent recap is a post-meeting experience, so it is not the fastest way to catch up during the live meeting. Reading the transcript manually is slower and does not use Copilot's summarization capability. The timing in the question is decisive: use in-meeting Copilot for immediate catch-up, and use Recap after the meeting.

• Question 74:

  You use the Researcher agent in Microsoft 365 Copilot to generate a report.

  What can you use to verify whether the report was generated by using valid sources?

  A. citations
  B. instructions
  C. memory
  D. capabilities
  A. citations

  ---

  Explanation

  Citations are the mechanism that lets a user inspect which sources supported a Copilot or Researcher response. A source-cited report should be checked by following the citations and confirming that the referenced material actually supports the claims. Instructions, memory, and capabilities can shape how an agent works, but they do not prove that a generated report used valid sources. Students should treat citations as the starting point for source verification, not as decoration.

• Question 75:

  HOTSPOT

  You use Microsoft 365 Copilot.

  For each of the following statements, select Yes if the statement is true. Otherwise, select No.

  NOTE: Each correct selection is worth one point.

  

  

  ---

  The selected responses are No, No, and Yes. Users can delete their own Copilot conversation history without needing an administrator to approve each deletion, although broader retention and compliance policies may still apply in the tenant.

  Deleting a conversation does not delete the SharePoint files that were referenced by that conversation, because those files are separate Microsoft 365 content. If an agent is uninstalled, saved prompts or related conversation records can remain available or manageable rather than being automatically destroyed with the agent.

• Question 76:

  You are explaining how Microsoft 365 Copilot protects organizational data.

  Which two statements are correct?

  Each correct answer presents part of the solution.

  NOTE: Each correct selection is worth one point.

  A. Copilot uses the signed-in user's existing Microsoft 365 permissions when grounding responses in work data.
  B. Copilot can summarize any file in the tenant if the prompt names the file exactly.
  C. Copilot removes sensitivity labels from generated content so users can edit it freely.
  D. Sensitivity labels, encryption, and SharePoint or OneDrive access controls can affect what Copilot can reference.
  E. Copilot grants temporary access to restricted content when a business justification is included in the prompt.
  A. Copilot uses the signed-in user's existing Microsoft 365 permissions when grounding responses in work data.
  D. Sensitivity labels, encryption, and SharePoint or OneDrive access controls can affect what Copilot can reference.

  ---

  Explanation

  Microsoft 365 Copilot uses the signed-in user's existing permissions when grounding answers in organizational data. Sensitivity labels, encryption, SharePoint permissions, and OneDrive permissions can all affect whether Copilot can access or use a piece of content. Copilot does not summarize any tenant file just because the prompt names it, and it does not remove labels or grant temporary access based on a business justification. The security model is permission-respecting, not tenant-wide by default.

• Question 77:

  HOTSPOT

  Select the answer that correctly completes the sentence.

  

  

  ---

  The selected completion says users accept the AI-generated output without verifying its accuracy. Overreliance is a human behavior risk, not a model attack or a data-exposure mechanism by itself. It happens when users treat generated content as correct without checking sources, reviewing assumptions, or applying professional judgment. The practical lesson is to verify important Copilot output before using it for business decisions or published material.

• Question 78:

  You are discussing Microsoft 365 Copilot with a colleague.

  The colleague asks which data Copilot uses to answer questions when using the Work scope.

  What should you tell your colleague?

  A. Copilot provides responses based only on data that the user can access and the general knowledge that Copilot was trained on.
  B. Copilot provides responses based on all the data in your organization's Microsoft 365 environment and the general knowledge that Copilot was trained on.
  C. Copilot provides responses based only on data that the user can access.
  D. Copilot provides responses based only on the general knowledge that Copilot was trained on.
  A. Copilot provides responses based only on data that the user can access and the general knowledge that Copilot was trained on.

  ---

  Explanation

  In Work scope, Microsoft 365 Copilot grounds responses in organizational data that the signed-in user is already allowed to access, and it can also use the model's general knowledge. It does not gain tenant-wide access merely because the prompt asks for information, and it does not ignore the user's Microsoft 365 permissions. A response based only on training knowledge would describe a web or general chat scenario rather than work-grounded Copilot. The main security rule is that Copilot respects the user's existing Microsoft 365 permissions when grounding in work data.

• Question 79:

  You are creating a custom agent in the Microsoft 365 Copilot app for the marketing team at your company.

  The agent will be used to produce marketing collateral, including copy, logos and artwork.

  What should you add to the agent?

  More than one answer choice may achieve the goal. Select the BEST answer.

  A. code interpreter
  B. image generator
  C. a suggested prompt
  D. a template
  B. image generator

  ---

  Explanation

  An image generator capability is the best fit when a marketing agent must produce logos, artwork, or other visual collateral. Code interpreter is for data analysis, calculations, and visualizations from data, not brand artwork creation. Suggested prompts and templates can guide users, but they do not add image-generation ability to the agent. Match the capability to the output type: image generator for creative visuals, code interpreter for computational analysis.

• Question 80:

  HOTSPOT

  You have a Microsoft 365 Copilot license.

  You plan to run several prompts.

  What should you use to ensure that Copilot generates the best response for each prompt?

  More than one answer choice may achieve the goal. To answer, select the BEST options in the answer area.

  NOTE: Each correct selection is worth one point.

  

  

  ---

  The selected mappings are Analyst for the spreadsheet feedback analysis, Researcher for the web and SharePoint brief, and Chat for the executive email. The first prompt asks for consolidating spreadsheets, sentiment analysis, and driver visualization, which is a data-analysis task for Analyst. The second prompt asks for reputable web sources, SharePoint content, citations, and synthesis, which fits Researcher. The third prompt asks for polished writing from an attached strategic plan, which ordinary Copilot Chat can handle without a specialized research or analytics agent.