Microsoft AI-103 Practice Test 21 - Prepaway
Practice Exams:

Pass Microsoft AI-103 Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers!

30 Days Free Updates, Instant Download!

Verified By Experts
Premium File
$89.99 $98.99

AI-103 Premium ETE File


  • Premium File 187 Questions & Answers. Last Update: Nov 15, 2026
Whats Included:
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates

Microsoft AI-103 Practice Test 21

Question 200

Situation: A utility operator wants to rotate service identities without redeploying hardcoded keys across several workloads. Which TWO proposed practices create material safety, correctness, or reliability risks? Select TWO answers.

  1. Grant all application identities owner access to the entire Azure subscription
  2. Authenticate the deployed application with its managed identity, assign narrowly scoped resource roles, and use the supported identity credential chain
  3. Embed administrator API keys in a container image shared across build environments
  4. The workload can invoke only its intended deployment and search store while a different principal and privileged write requests are denied
  5. Inspect records that explain the affected process, then check the result described in this case: A utility operator wants to rotate service identities without redeploying hardcoded keys across several workloads

Explanation

Correct Answer

A:

Because a utility operator wants to rotate service identities without redeploying hardcoded keys across several workloads, this grants access or authority beyond the scope established in the case. “Grant all application identities owner access to the entire Azure subscription” would violate the requirement. Service authentication and data authorization are separate checks; possession of a Foundry project URL does not grant deployment or knowledge-store permissions.

C:

A utility operator wants to rotate service identities without redeploying hardcoded keys across several workloads. “Embed administrator API keys in a container image shared across build environments” would violate the requirement. Service authentication and data authorization are separate checks; possession of a Foundry project URL does not grant deployment or knowledge-store permissions.

Incorrect Answers

B:

“Authenticate the deployed application with its managed identity, assign narrowly scoped resource roles, and use the supported identity credential chain” is the controlled implementation, not an unsafe workaround. Service authentication and data authorization are separate checks; possession of a Foundry project URL does not grant deployment or knowledge-store permissions.

D:

Because a utility operator wants to rotate service identities without redeploying hardcoded keys across several workloads, “The workload can invoke only its intended deployment and search store while a different principal and privileged write requests are denied” is the required outcome of a working control, not an action that weakens it. Service authentication and data authorization are separate checks; possession of a Foundry project URL does not grant deployment or knowledge-store permissions. Confirm that the workload can invoke only its intended deployment and search store while a different principal and privileged write requests are denied.

E:

“Inspect records that explain the affected process, then check the result described in this case: A utility operator wants to rotate service identities without redeploying hardcoded keys across several” adds a relevant safeguard or verification step rather than weakening the solution. Service authentication and data authorization are separate checks; possession of a Foundry project URL does not grant deployment or knowledge-store permissions.

Question 201

Architecture case: A children-facing tutor must reject self-harm instructions while still answering ordinary biology questions. What configuration should the engineer choose for this workload?

  1. Discard challenge prompts after a single policy change
  2. Apply the proposed setting to all connected workloads before testing the case-specific boundary: a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions
  3. Run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds
  4. Assume every blocked request is inherently an attack

Explanation

Correct Answer

C:

Because a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions, a green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

Incorrect Answers

A:

“Discard challenge prompts after a single policy change” does not implement the control that the facts call for. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. Instead, run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds.

B:

“Apply the proposed setting to all connected workloads before testing the case-specific boundary: a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions” does not implement the control that the facts call for. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. Instead, run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds.

D:

Because a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions, “Assume every blocked request is inherently an attack” would not meet the requirement. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. Instead, run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds.

Question 202

Production requirement: A children-facing tutor must reject self-harm instructions while still answering ordinary biology questions. What technical fact prevents the simplest shortcut from being reliable?

  1. The fact that the team started work is enough to prove correctness for this exact case: a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions
  2. The most powerful or least restricted model is always the safest choice regardless of task conditions
  3. Successful provisioning by itself proves all security, output, and operational constraints are satisfied
  4. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses

Explanation

Correct Answer

D:

Because a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions, a green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. Run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds. The decisive evidence is that measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

Incorrect Answers

A:

“The fact that the team started work is enough to prove correctness for this exact case: a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions” does not fully explain the decisive technical boundary in this case. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

B:

“The most powerful or least restricted model is always the safest choice regardless of task conditions” does not fully explain the decisive technical boundary in this case. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

C:

A children-facing tutor must reject self-harm instructions while still answering ordinary biology questions. “Successful provisioning by itself proves all security, output, and operational constraints are satisfied” does not fully explain the decisive technical boundary in this case. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. The implementation must run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds.

Question 203

A team is investigating this task: A children-facing tutor must reject self-harm instructions while still answering ordinary biology questions. What end-to-end test would distinguish successful operation from an installed configuration?

  1. The operator reports success without reproducing the observed behavior: a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions
  2. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates
  3. The team closes a change ticket without replaying a representative positive and negative test
  4. A project dashboard remains green, although the affected request, output and permissions were not checked

Explanation

Correct Answer

B:

A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

Incorrect Answers

A:

“The operator reports success without reproducing the observed behavior: a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions” alone does not establish all the required behavior. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

C:

“The team closes a change ticket without replaying a representative positive and negative test” alone does not establish all the required behavior. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

D:

A children-facing tutor must reject self-harm instructions while still answering ordinary biology questions. “A project dashboard remains green, although the affected request, output and permissions were not checked” would not meet the requirement. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

Question 204

Situation: A children-facing tutor must reject self-harm instructions while still answering ordinary biology questions. Which diagnostic investigation best isolates the issue before another change?

  1. Assume the general service is at fault without checking the affected operation in this case: a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions
  2. Restart every affected component before collecting request or tool traces
  3. Read only the broad service-health summary and skip actual workload evidence
  4. Review harmful and benign evaluation slices, moderation category severity, blocked-request logs, appeals and versioned policy decisions

Explanation

Correct Answer

D:

Because a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions, review harmful and benign evaluation slices, moderation category severity, blocked-request logs, appeals and versioned policy decisions. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

Incorrect Answers

A:

“Assume the general service is at fault without checking the affected operation in this case: a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions” would not meet the requirement. Review harmful and benign evaluation slices, moderation category severity, blocked-request logs, appeals and versioned policy decisions.

B:

Because a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions, “Restart every affected component before collecting request or tool traces” would not be the most direct first check for this failure. Review harmful and benign evaluation slices, moderation category severity, blocked-request logs, appeals and versioned policy decisions. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

C:

“Read only the broad service-health summary and skip actual workload evidence” would not meet the requirement. Review harmful and benign evaluation slices, moderation category severity, blocked-request logs, appeals and versioned policy decisions.

Question 205

Architecture case: A children-facing tutor must reject self-harm instructions while still answering ordinary biology questions. Which TWO shortcuts should be rejected for this scenario? Select TWO answers.

  1. Inspect records that explain the affected process, then check the result described in this case: A children-facing tutor must reject self-harm instructions while still answering ordinary biology questions
  2. Turn off safety filtering because ordinary users report occasional refusals
  3. Approve release using only general task accuracy without abuse-case tests
  4. Run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds
  5. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates

Explanation

Correct Answer

B:

“Turn off safety filtering because ordinary users report occasional refusals” would violate the requirement. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

C:

Because a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions, “Approve release using only general task accuracy without abuse-case tests” would violate the requirement. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

Incorrect Answers

A:

“Inspect records that explain the affected process, then check the result described in this case: A children-facing tutor must reject self-harm instructions while still answering ordinary biology questions” adds a relevant safeguard or verification step rather than weakening the solution. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

D:

Because a children-facing tutor must reject self-harm instructions while still answering ordinary biology questions, “Run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds” is the controlled implementation, not an unsafe workaround. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. Confirm that measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

E:

“Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates” is the required outcome of a working control, not an action that weakens it. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. Confirm that measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

Question 206

Production requirement: A bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions. Which next change most directly addresses the reported technical risk?

  1. Discard challenge prompts after a single policy change
  2. Run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds
  3. Apply the proposed setting to all connected workloads before testing the case-specific boundary: a bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions
  4. Assume every blocked request is inherently an attack

Explanation

Correct Answer

B:

Run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

Incorrect Answers

A:

“Discard challenge prompts after a single policy change” does not implement the control that the facts call for. Instead, run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds.

C:

“Apply the proposed setting to all connected workloads before testing the case-specific boundary: a bank assistant receives malicious instructions embedded in a retrieved customer-support article” does not implement the control that the facts call for. Instead, run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds.

D:

“Assume every blocked request is inherently an attack” would not meet the requirement. Instead, run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds.

Question 207

A team is investigating this task: A bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions. Which explanation best captures the key behavior of this architecture?

  1. The most powerful or least restricted model is always the safest choice regardless of task conditions
  2. Successful provisioning by itself proves all security, output, and operational constraints are satisfied
  3. The fact that the team started work is enough to prove correctness for this exact case: a bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions
  4. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses

Explanation

Correct Answer

D:

A bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

Incorrect Answers

A:

“The most powerful or least restricted model is always the safest choice regardless of task conditions” would not be the most direct first check for this failure. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

B:

“Successful provisioning by itself proves all security, output, and operational constraints are satisfied” would not be the most direct first check for this failure. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

C:

“The fact that the team started work is enough to prove correctness for this exact case: a bank assistant receives malicious instructions embedded in a retrieved customer-support article” would not be the most direct first check for this failure. A green ordinary-task benchmark does not establish safety against harmful inputs, indirect prompt injection or domain-specific prohibited responses.

Question 208

Situation: A bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions. Which result would demonstrate that the proposed remedy works end to end?

  1. A project dashboard remains green, although the affected request, output and permissions were not checked
  2. The team closes a change ticket without replaying a representative positive and negative test
  3. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates
  4. The operator reports success without reproducing the observed behavior: a bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions

Explanation

Correct Answer

C:

A bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates. Run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds.

Incorrect Answers

A:

“A project dashboard remains green, although the affected request, output and permissions were not checked” would not meet the requirement. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

B:

Because a bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions, “The team closes a change ticket without replaying a representative positive and negative test” alone does not establish all the required behavior. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

D:

“The operator reports success without reproducing the observed behavior: a bank assistant receives malicious instructions embedded in a retrieved customer-support article” alone does not establish all the required behavior. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

Question 209

Architecture case: A bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions. What evidence should the engineer correlate to verify the actual failure cause?

  1. Assume the general service is at fault without checking the affected operation in this case: a bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions
  2. Restart every affected component before collecting request or tool traces
  3. Review harmful and benign evaluation slices, moderation category severity, blocked-request logs, appeals and versioned policy decisions
  4. Read only the broad service-health summary and skip actual workload evidence

Explanation

Correct Answer

C:

Because a bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions, measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates. Run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds.

Incorrect Answers

A:

“Assume the general service is at fault without checking the affected operation in this case: a bank assistant receives malicious instructions embedded in a retrieved customer-support article” would not meet the requirement. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

B:

“Restart every affected component before collecting request or tool traces” alone does not establish all the required behavior. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

D:

“Read only the broad service-health summary and skip actual workload evidence” would not meet the requirement. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

Question 210

Production requirement: A bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions. Which TWO decisions would undermine the required system boundary? Select TWO answers.

  1. Run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds
  2. Approve release using only general task accuracy without abuse-case tests
  3. Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates
  4. Turn off safety filtering because ordinary users report occasional refusals
  5. Inspect records that explain the affected process, then check the result described in this case: A bank assistant receives malicious instructions embedded in a retrieved customer-support article and must not execute external actions

Explanation

Correct Answer

B:

“Approve release using only general task accuracy without abuse-case tests” would violate the requirement.

D:

“Turn off safety filtering because ordinary users report occasional refusals” would violate the requirement.

Incorrect Answers

A:

“Run a documented abuse-case evaluation set, tune filters and prompt protections, and gate release on measurable false-positive and false-negative thresholds” is the controlled implementation, not an unsafe workaround. Confirm that measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates.

C:

“Measured safety violations fall below agreed thresholds while legitimate requests retain acceptable completion and escalation rates” is the required outcome of a working control, not an action that weakens it.

E:

“Inspect records that explain the affected process, then check the result described in this case: A bank assistant receives malicious instructions embedded in a retrieved customer-support article” adds a relevant safeguard or verification step rather than weakening the solution.

Developing AI Apps and Agents on Azure (AI-103) Practice Tests