Free Practice Questions for Microsoft AB-730 Exam

Microsoft 365 Copilot can be accessed via its dedicated web experience for enterprise users. The correct web entry point for Microsoft 365 Copilot is https://m365.cloud.microsoft , which provides authenticated access to Copilot features within the Microsoft 365 environment.

This URL routes users into the Microsoft 365 ecosystem where Copilot integrates with organizational data such as SharePoint, OneDrive, Teams, and Outlook. It ensures that users are authenticated through Microsoft Entra ID and that access controls are enforced according to tenant policies.

Option B refers to Copilot Studio, which is used to build and manage custom copilots and agents rather than access the Microsoft 365 Copilot chat experience. Option C (My Apps) is a general application launcher portal. Option D is an incorrect account management URL and does not provide Copilot access.

Therefore, to access Microsoft 365 Copilot from a web browser, the correct URL is https://m365.cloud.microsoft .

Microsoft 365 Copilot follows Microsoft’s Responsible AI principles and enforces strict content safety policies. When a prompt violates safety guidelines—such as containing harmful, abusive, illegal, or restricted content—the system may refuse to generate a response. The refusal message shown is consistent with safety filtering behavior.

Generative AI systems include moderation layers that evaluate prompts before generating output. If the prompt is classified as unsafe or non-compliant with policy, Copilot blocks the request and encourages the user to try a different topic.

A vague prompt typically results in a generic or clarifying response rather than a refusal. There is no fixed limit of five requests per prompt. Exceeding the context window usually results in truncation or processing errors, not a safety-based refusal message.

Therefore, the most likely cause of the response is that the prompt contains language that violates safety guidelines.

==================================================

Microsoft 365 Copilot operates within the security, compliance, and identity boundaries of a Microsoft 365 tenant. Shared prompts, prompt links, and Copilot artifacts are governed by organizational access controls and tenant isolation. If a prompt is created and shared from outside your organization, cross-tenant access may not be supported depending on the sharing configuration and administrative policies.

When a user attempts to open a prompt that resides in another organization’s tenant without proper cross-tenant sharing permissions, Copilot cannot locate or validate the resource within the user’s own environment. As a result, the system displays a “Prompt not found” message.

Option B would typically result in an access or permissions error rather than the prompt being unavailable entirely. Sensitivity labels and scheduled prompts do not inherently cause a “not found” error. Therefore, the most likely cause is that the prompt exists outside your organization’s tenant boundary and is not accessible to you.

==================================================

 When Microsoft 365 Copilot searches for content on the web, it uses the Bing search service.

Answer: Yes

 When Microsoft 365 Copilot searches for content on the web, it uploads the entire conversation history into the search request.

Answer: No

 When Microsoft 365 Copilot searches for content on the web, the query is anonymized and stripped of any user-identifiable information.

Answer: Yes

Microsoft 365 Copilot can use web grounding when a response requires current or public information. For web search, Copilot uses the Bing search service to retrieve relevant web results, so Statement 1 is Yes . This allows Copilot to supplement its response with up-to-date internet content when web grounding is enabled or selected.

Statement 2 is No because Copilot does not send the entire conversation history to Bing as a web search request. Instead, Copilot generates a search query based on the user’s prompt and relevant context. Sending the complete conversation would create unnecessary privacy and data exposure risk, which would conflict with Microsoft’s enterprise data protection model.

Statement 3 is Yes because Microsoft 365 Copilot is designed to protect user and tenant information during web grounding. Search queries are handled in a privacy-preserving way, with user-identifiable and tenant-identifiable information removed before the query is sent to Bing.

Microsoft 365 Copilot agents dynamically reflect updates such as new suggested prompts. When an agent is updated, users who already have the agent installed do not need to reinstall or have the agent reshared. Instead, the updated configuration becomes available when the agent is used again.

Suggested prompts are surfaced when a user interacts with the agent. Therefore, once Ben Smith issues any prompt to Agent1, the agent session refreshes and displays the latest suggested prompts, including the newly added one.

Stopping and resharing the agent (Option A) or reinstalling it (Option B) is unnecessary overhead. Option D refers to a non-standard command and is not part of Microsoft Copilot functionality.

Thus, the correct action is simply to have the user interact with the agent again.

==================================================

The first statement is correct because Microsoft 365 provides transparency and user control through the My Account portal, where users can review their Copilot activity history. This aligns with Microsoft’s responsible AI and data governance principles, ensuring visibility into AI interactions.

The second statement is incorrect. Deleting Copilot activity history removes stored interaction records (such as prompts and responses), but it does not automatically delete associated notebooks, documents, or pages stored in services like OneDrive or SharePoint. Those files remain governed by standard Microsoft 365 retention and lifecycle policies.

The third statement is correct because users can delete their entire Copilot activity history, including both prompts and generated responses. This reinforces enterprise-grade privacy controls and regulatory compliance requirements.

These controls demonstrate core generative AI fundamentals: transparency, user data ownership, security boundaries, and responsible lifecycle management of AI-generated interactions within Microsoft 365 environments.

Prompt-grounding means ensuring Copilot’s response is based on authoritative, relevant source material rather than inferred or generalized information. Microsoft guidance for Copilot emphasizes that the most effective way to ground responses is to explicitly point Copilot to the content you want it to use—such as specific SharePoint documents, OneDrive files, emails, meeting notes, or other Microsoft 365 resources the user has permission to access.

Option D is therefore the best answer because referencing specific files anchors Copilot’s reasoning to the exact material that should be used for facts, terminology, and details. This reduces hallucinations (fabrication) and improves accuracy and traceability, especially when Copilot can provide citations back to the referenced sources.

Option B (desired output format) improves structure and readability but does not, by itself, ground content. Option A can sometimes improve clarity, but shorter prompts do not guarantee grounding. Option C may help comprehension in some cases, but it is not a grounding strategy and can be counterproductive if the source material is technical.

Therefore, the best approach to improve prompt-grounding is to reference specific files .

All three statements are false because generative AI responses are not guaranteed to be identical even when the prompt is the same. First, when grounded in the web , results can vary due to changing web content, different retrieved sources, or differences in how information is summarized at run time. Second, when grounded in your organization’s data , responses can change based on updates to files, emails, meetings, permissions, or which specific items Copilot retrieves as the most relevant context at that moment. Third, even when relying only on the model’s general knowledge, large language models are probabilistic: they may choose different wording, structure, examples, or emphasis across runs, especially when temperature/decoding settings and internal routing differ. In business scenarios, this means Copilot outputs should be treated as drafts that may require validation, and repeatability should be improved by adding precise constraints (cite specific sources, use fixed formats, specify exact sections, and request verbatim quotes where appropriate).

This scenario is an example of fabrication, which is commonly referred to in generative AI contexts as a hallucination. Fabrication occurs when an AI system generates information that appears credible but is factually incorrect, invented, or unsupported by verifiable sources.

According to Microsoft AI Business Professional guidance, large language models predict text based on patterns learned during training. They do not “know” facts in the human sense. As a result, when asked to generate reports using web-based information, the model may produce plausible-sounding but fictional details if sufficient grounding or reliable sources are not provided.

Deepfake refers specifically to synthetic media such as manipulated images, audio, or video. Overreliance describes a human behavior risk where users trust AI outputs without verification. Prompt injection is a malicious technique designed to manipulate model behavior. Bias refers to systematic unfairness in outputs.

In this case, the presence of fictional information in the generated report directly aligns with fabrication, making option B the correct answer.