mediumMultiple SelectObjective-mapped
Two Actions for Responsible AI with Copilot in Dynamics 365
A Dynamics 365 administrator is configuring Copilot features for the sales team. Which TWO actions should the administrator take to ensure responsible AI usage?
Quick Answer
The correct answer is to review and test Copilot responses in a sandbox environment before enabling for all users, and to enable the feedback mechanism so users can report inaccurate or inappropriate outputs. These two actions are essential because responsible AI usage with Copilot configuration requires both proactive validation of AI behavior and a continuous feedback loop to catch unforeseen issues. Testing in a sandbox ensures the AI’s suggestions align with business rules and data privacy before broad rollout, while the feedback mechanism directly supports Microsoft’s responsible AI principles of transparency and accountability. On the MB-920 exam, this question tests your understanding that responsible AI isn’t just about technical accuracy—it’s about governance and user empowerment. A common trap is choosing only technical controls like data filtering, but the exam emphasizes human oversight and reporting. Remember the mnemonic “Sandbox and Speak Up”: sandbox for pre-deployment safety, and speak up via feedback to catch post-deployment problems.
⚠ Common exam trap
Candidates often confuse general security best practices (like encryption) or localization settings with the specific requirements for responsible AI governance, which center on transparency, accountability, and user feedback.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Enable the feedback mechanism to allow users to report inaccurate responses
Enabling the feedback mechanism allows users to report inaccurate or inappropriate Copilot responses, which is a core requirement for responsible AI usage. This feedback loop helps administrators identify and correct issues, ensuring the AI system remains trustworthy and compliant with Microsoft's responsible AI principles. Without this mechanism, the organization would lack visibility into potential harms or inaccuracies generated by the AI.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Enable the feedback mechanism to allow users to report inaccurate responses
Why this is correct
Feedback helps Microsoft improve the AI and address issues.
- ✗
Disable access to all customer data to prevent Copilot from using it
Why it's wrong here
Copilot needs data to function; disabling access would break it.
- ✗
Restrict Copilot to only respond in English
Why it's wrong here
Language restriction is not a responsible AI practice; it's a localization setting.
- ✓
Review and test Copilot responses in a sandbox environment before enabling for all users
Why this is correct
This ensures the AI behaves as expected in the organization's context.
- ✗
Encrypt all data at rest and in transit
Why it's wrong here
While important for security, this is not specific to responsible AI usage.
Go deeper
Related to this question
About these practice questions
Courseiva writes every MB-920 question from scratch — 930 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
Same concept, more angles
2 more ways this is tested on MB-920
These questions test the same concept from different angles. Work through them to make sure you can recognise it however the exam phrases it.
Variation 1. A Dynamics 365 Customer Service agent is using Copilot to draft email responses to customers. The agent notices that Copilot sometimes suggests responses that include incorrect information about the company's return policy. What should the agent do to ensure accurate responses?
medium- A.Disable Copilot and draft all responses manually
- ✓ B.Edit the suggested response to correct the inaccuracies before sending
- C.Ask the administrator to change the Copilot language model
- D.Provide feedback to Copilot by rating the response as unhelpful
Why B: Copilot in Dynamics 365 Customer Service generates draft responses based on available data, but it may not always reflect the most current or accurate policy details. The agent is responsible for reviewing and editing the suggested content before sending, ensuring the final response is accurate and compliant with company policies. This aligns with the principle that Copilot is an assistive tool, not a replacement for human judgment.
Variation 2. A company is implementing Dynamics 365 Business Central and wants to use Copilot for generating marketing text for product descriptions. However, the generated text often contains inaccuracies about product specifications. What should the administrator do to improve the accuracy of Copilot-generated content?
hard- A.Disable Copilot for marketing text and rely on manual descriptions
- B.Change the Copilot language model to a more specialized one
- ✓ C.Provide high-quality, well-structured product data in the system
- D.Regularly review and edit the generated text so Copilot learns from corrections
Why C: Copilot in Dynamics 365 Business Central generates marketing text by leveraging existing product data stored in the system. If the product specifications are inaccurate or incomplete, the generated content will reflect those flaws. Providing high-quality, well-structured product data ensures Copilot has accurate source information to work with, directly improving the reliability of its output.
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This MB-920 practice question is part of Courseiva's free Microsoft certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the MB-920 exam.