https://kxbjsyuhceggsyvxdkof.supabase.co/storage/v1/object/public/file-images/Microsoft_AB-730/page_39_img_2.jpg

https://kxbjsyuhceggsyvxdkof.supabase.co/storage/v1/object/public/file-images/Microsoft_AB-730/page_27_img_1.jpg

https://kxbjsyuhceggsyvxdkof.supabase.co/storage/v1/object/public/file-images/Microsoft_AB-730/page_20_img_1.jpg

Prompt injection is a generative AI security risk where an attacker inserts instructions (often hidden

in text, documents, webpages, or user inputs) to override or manipulate the assistant’s intended

behavior. This can lead to unintended actions such as ignoring policy controls, producing unsafe

outputs, or attempting to reveal sensitive information. Because generative AI systems follow natural-

language instructions, they can be socially engineered to prioritize malicious content unless

safeguards are in place. This is why prompt injection can cause data exposure (for example,

attempting to extract confidential content from grounded sources) and can also embed harmful

instructions that redirect the model’s behavior. In enterprise settings like Microsoft 365 Copilot,

mitigations include grounding boundaries, permission trimming, content filtering, and instruction

hierarchy (system policies over user instructions). From a business governance perspective, users

should treat untrusted inputs (emails, documents, web text) as potentially hostile and apply least-

privilege access and validation when using AI outputs in decision-making.

https://kxbjsyuhceggsyvxdkof.supabase.co/storage/v1/object/public/file-images/Microsoft_AB-730/page_9_img_1.jpg

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.

Microsoft 365 Copilot in Teams is designed to provide real-time meeting assistance, including

summarizing discussions, identifying key decisions, and highlighting action items. When joining a

meeting late, the most efficient and immediate method to catch up is to directly prompt Copilot

within the Teams meeting experience.

Option A is the best answer because Copilot in Teams can summarize what has occurred so far during

the live meeting. It leverages the meeting transcript, speaker attribution, and contextual signals to

generate a concise, structured summary in real time. This minimizes delay and provides actionable

insights instantly.

Reading the transcript (Option B) would require manual review and is not the fastest method.

Intelligent recap (Option D) is typically available after the meeting concludes and processes recording

data. Asking Copilot from the separate Microsoft 365 Copilot app (Option C) may not provide

immediate, in-meeting contextual awareness.

Therefore, the fastest and most effective approach is to ask Copilot in Teams to summarize what you

missed.

Microsoft 365 Copilot is integrated with Microsoft Search to ground its responses in your organization's data. To find existing documents, such as similar training plans, you would prompt Copilot to perform a search. Copilot then uses the Microsoft Search index to locate relevant files, emails, and other content across Microsoft 365 services that you have permission to access. This allows you to leverage existing organizational knowledge directly within your workflow.

B. Designer is a feature for creating visually appealing content and layouts, not for searching for existing documents within your organization.

C. Apps, in the context of Copilot, refer to plugins or connectors that extend its capabilities to third-party services, not the core function for finding internal files.

D. While you might search for content on SharePoint "Pages," the fundamental and overarching capability used to find any type of document is "Search."

1. Microsoft Learn. (2024). Overview of Copilot for Microsoft 365. "How Copilot for Microsoft 365 works". This section explains that Copilot processes a prompt by making a call to Microsoft Graph and Microsoft Search to retrieve business content.

2. Microsoft Learn. (2024). Data, Privacy, and Security for Microsoft Copilot for Microsoft 365. "Copilot for Microsoft 365 and your data". This document details how Copilot uses Microsoft Search to access organizational content while respecting user permissions.

The generation of fictional or factually incorrect information by a large language model is known as fabrication or hallucination. This occurs because the model is designed to predict the next most likely sequence of words to form a coherent response, rather than retrieving facts from a knowledge base. As a result, it can confidently present information that is not grounded in the source data or reality. This is a recognized limitation of current generative AI, highlighting the need for human verification of AI-generated content.

A. A deepfake is a synthetically altered video or audio recording, not a text-based response from an AI model.

C. Overreliance is a human behavior of trusting AI output without critical evaluation, not the AI's act of generating false information.

D. Prompt injection is a security exploit where a user tricks an AI into bypassing its safety constraints, which is not what is described.

E. Bias refers to skewed or prejudiced outputs based on the training data, whereas fabrication is the creation of entirely new, untrue information.

1. Microsoft. (2023). Microsoft Responsible AI Standard, v2. Section 2.2, "Reliability and Safety," discusses the need to mitigate "the risk of the system generating content that is not factually accurate (often referred to as 'hallucinations' or 'fabrications')."

2. Stanford University Human-Centered Artificial Intelligence (HAI). (2023). On the Opportunities and Risks of Foundation Models. Section 4.2, "Misinformation," discusses how models can "generate plausible-sounding but false or nonsensical text," a phenomenon referred to as hallucination or fabrication.

To ensure transparency and allow users to verify the information it generates, Microsoft 365 Copilot provides citations. When Copilot synthesizes information from your organization's data or public sources to create content, it includes references pointing back to the original source documents, emails, or web pages. By reviewing these citations, a user can directly access the source material to confirm its validity, context, and accuracy, which is a core principle of responsible AI implementation.

B. Instructions are the user's prompt to the AI; they do not verify the sources the AI used to generate the response.

C. Memory refers to the conversational context the AI maintains, not the external data sources it references for its answers.

D. Capabilities describe the agent's functions (e.g., web search), not the specific sources used for a particular output.

1. Microsoft. (2024). How Microsoft Copilot for Microsoft 365 works. Microsoft Learn. Retrieved from https://learn.microsoft.com/en-us/microsoft-365-copilot/how-microsoft-365-copilot-works. (Under the "How Copilot creates a response" section, it states, "Copilot provides citations to the source material it used, so you can verify the responses.").

2. Microsoft. (2024). Data, Privacy, and Security for Microsoft Copilot for Microsoft 365. Microsoft Learn. Retrieved from https://learn.microsoft.com/en-us/microsoft-365-copilot/microsoft-365-copilot-privacy. (This document emphasizes that Copilot responses are "grounded" in the user's data and that citations are provided for traceability).

The URL https://m365.cloud.microsoft is a modern endpoint that directs users to the central Microsoft 365 web portal (microsoft365.com). This portal is a primary access point for the integrated Microsoft 365 Copilot experience, alongside web versions of apps like Word, Excel, and Teams. While other URLs exist for specific Copilot-related services, this URL is the correct entry point for the core Microsoft 365 web experience where Copilot is featured.

B. This URL is for Microsoft Copilot Studio, a low-code platform for building and customizing copilots, not for end-user access.

C. This is the My Apps portal, which lists a user's entitled enterprise applications but is not the direct service URL for Copilot.

D. This URL is for managing a user's Microsoft account profile and security settings, not for accessing applications.

1. Microsoft Learn. (2024). Overview of Microsoft 365 Copilot. "Users can access Copilot in Microsoft 365 at microsoft365.com, and in the apps they use every day..." Document ID: M365-Copilot-Overview.

2. Microsoft 365 Documentation. (2024). Where can I get Microsoft Copilot?. "Copilot is available... on the web at Copilot.microsoft.com and at microsoft365.com." Section: "Copilot in Microsoft 365".

To simplify referencing the same files, a user can leverage two distinct methods. First, the Notebook feature within the main Copilot interface allows for drafting, editing, and saving complex prompts that can reference specific files, making it easy to reuse the context without re-uploading. Second, for a more robust and permanent solution, the user can use Microsoft Copilot Studio to create a custom agent (a topic-specific copilot). By uploading the five files as a knowledge source to this agent, they become a persistent part of its grounding data, accessible in any conversation with that agent.

A. Saving a prompt in the Prompt Gallery saves the prompt text but does not retain any file attachments.

C. Fine-tuning a large language model is a highly complex, developer-centric task for altering core model behavior, not for simple file referencing.

E. Zipping files streamlines the upload process into one action but still requires the user to manually upload the file in each new chat.

1. Microsoft Learn. (2024). Microsoft Copilot Studio - Use documents as a data source. "You can use documents to have your copilot generate answers from... The content from the documents is used by your copilot to generate answers." Section: "Upload a document".

2. Microsoft Copilot Documentation. (2024). Using the Notebook interface. "Notebook is a new interface... that lets you iterate on your prompts. It's a great place to refine your prompts and ensure you get the best possible response." This interface is used to maintain context, including file references, for reuse.