Practice Exams:

Microsoft AI-103 Dumps

EXAM: Developing AI Apps and Agents on Azure
Verified By Experts
Premium Bundle
$49.99

AI-103 Premium ETE File


  • Premium File 187 Questions & Answers
Whats Included:
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates

Microsoft AI-103 Practice Test 25

Question 240

During a operational readiness review, a compliance specialist must accept a change for sensitive content disclosure. The issue to be addressed is: An agent answer reveals confidential retrieved data to a user lacking authorization. Which concrete verification action demonstrates that the relevant behavior works?

  1. Query as an unauthorized user and confirm no private snippets or citations leak
  2. Mask only the final text while retrieving full private documents under an overprivileged tool
  3. Present conflicting guidance and confirm approval is requested instead of refund execution
  4. Apply per-user data permissions before retrieval and response generation

Explanation

Correct Answer

A:

To verify this change, the software consultancy should query as an unauthorized user and confirm no private snippets or citations leak. Operators still need the implementation to apply per-user data permissions before retrieval and response generation and the preventive requirement to audit response-grounding sources against requester entitlements.

Incorrect Answers

B:

In the software consultancy acceptance review, trying to mask only the final text while retrieving full private documents under an overprivileged tool does not test the exact sensitive content disclosure condition where an agent answer reveals confidential retrieved data to a user lacking authorization; the relevant test is to query as an unauthorized user and confirm no private snippets or citations leak. Acceptance should instead demonstrate that the service can query as an unauthorized user and confirm no private snippets or citations leak, using the described failure as the test criterion.

C:

For human escalation after uncertainty, it can be sensible to present conflicting guidance and confirm approval is requested instead of refund execution if an agent cannot reliably interpret conflicting refund policies. Without the actual change to query as an unauthorized user and confirm no private snippets or citations leak, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

D:

The action to apply per-user data permissions before retrieval and response generation can contribute to sensitive content disclosure as a implementation step; here, an agent answer reveals confidential retrieved data to a user lacking authorization. In the software consultancy acceptance review, trying to apply per-user data permissions before retrieval and response generation does not test the exact sensitive content disclosure condition where an agent answer reveals confidential retrieved data to a user lacking authorization; the relevant test is to query as an unauthorized user and confirm no private snippets or citations leak.

Question 241

At a energy cooperative, the product support lead is preparing a cost review. The concrete requirement is: A team cannot identify whether licensed data is allowed in its evaluation set. Which design action BEST addresses the stated need?

  1. Capture dataset provenance and permitted use rights before ingestion
  2. Assume every publicly reachable webpage is licensed for unrestricted reuse
  3. Require evidence-backed uncertainty and human approval before adverse decisions
  4. Trace random sample records back to permitted originals and terms

Explanation

Correct Answer

A:

The proposed implementation is to capture dataset provenance and permitted use rights before ingestion; this matches the operational requirement that a team cannot identify whether licensed data is allowed in its evaluation set. The team can trace random sample records back to permitted originals and terms. The team should also record source permission status and evaluation dataset revision history.

Incorrect Answers

B:

Choosing to assume every publicly reachable webpage is licensed for unrestricted reuse cannot directly replace the specified action to capture dataset provenance and permitted use rights before ingestion when a team cannot identify whether licensed data is allowed in its evaluation set. The actual implementation must capture dataset provenance and permitted use rights before ingestion, because the stated service condition is that a team cannot identify whether licensed data is allowed in its evaluation set.

C:

Choosing to require evidence-backed uncertainty and human approval before adverse decisions would direct effort toward false accusation risk, whose characteristic issue is that an incident triage agent accuses employees based on unverified telemetry. The relevant corrective or preventive action here remains to capture dataset provenance and permitted use rights before ingestion, with its own validation criteria.

D:

The action to trace random sample records back to permitted originals and terms can contribute to training data provenance as a targeted acceptance check step; here, a team cannot identify whether licensed data is allowed in its evaluation set. Choosing to trace random sample records back to permitted originals and terms cannot directly replace the specified action to capture dataset provenance and permitted use rights before ingestion when a team cannot identify whether licensed data is allowed in its evaluation set.

Question 242

After a reliability exercise at a industrial maintenance team, the team observes this failure: A team cannot identify whether licensed data is allowed in its evaluation set. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Trace random sample records back to permitted originals and terms
  2. Assume every publicly reachable webpage is licensed for unrestricted reuse
  3. Disable all safety filtering in production to eliminate benign rejections
  4. Record source permission status and evaluation dataset revision history
  5. Capture dataset provenance and permitted use rights before ingestion

Explanation

Correct Answer

B:

This design mistake is to assume every publicly reachable webpage is licensed for unrestricted reuse; it conflicts with the actual training data provenance behavior needed when a team cannot identify whether licensed data is allowed in its evaluation set. The industrial maintenance team should instead capture dataset provenance and permitted use rights before ingestion, followed by a check to trace random sample records back to permitted originals and terms.

Incorrect Answers

A:

The alternative to trace random sample records back to permitted originals and terms is a targeted acceptance check step rather than the faulty assumption being diagnosed. This industrial maintenance team failure-analysis alternative of trace random sample records back to permitted originals and terms does not diagnose the particular training data provenance defect where a team cannot identify whether licensed data is allowed in its evaluation set; the keyed mistake is assume every publicly reachable webpage is licensed for unrestricted reuse.

C:

This approach relates to content filter false positives: a clinical educational assistant blocks harmless health discussions. If the team chose it instead, the targeted check to trace random sample records back to permitted originals and terms could still fail because the underlying required action is to assume every publicly reachable webpage is licensed for unrestricted reuse.

D:

The alternative to record source permission status and evaluation dataset revision history is a ongoing preventive rule step rather than the faulty assumption being diagnosed. This industrial maintenance team failure-analysis alternative of record source permission status and evaluation dataset revision history does not diagnose the particular training data provenance defect where a team cannot identify whether licensed data is allowed in its evaluation set; the keyed mistake is assume every publicly reachable webpage is licensed for unrestricted reuse.

E:

The alternative to capture dataset provenance and permitted use rights before ingestion is a implementation step rather than the faulty assumption being diagnosed. This industrial maintenance team failure-analysis alternative of capture dataset provenance and permitted use rights before ingestion does not diagnose the particular training data provenance defect where a team cannot identify whether licensed data is allowed in its evaluation set; the keyed mistake is assume every publicly reachable webpage is licensed for unrestricted reuse.

Question 243

The research university is responding to a source update. A service administrator discovers this issue: A team cannot identify whether licensed data is allowed in its evaluation set. What corrective step should the team take to restore the intended functionality most directly?

  1. Trace random sample records back to permitted originals and terms
  2. Revoke excess access and invalidate cached sensitive retrieval results
  3. Assume every publicly reachable webpage is licensed for unrestricted reuse
  4. Remove unsupported materials and replace them with authorized examples

Explanation

Correct Answer

D:

The direct repair is to remove unsupported materials and replace them with authorized examples. Once the condition is corrected, the research university should trace random sample records back to permitted originals and terms. To reduce recurrence, it should also record source permission status and evaluation dataset revision history; those steps provide different recovery and prevention functions.

Incorrect Answers

A:

The action to trace random sample records back to permitted originals and terms can contribute to training data provenance as a targeted acceptance check step; here, a team cannot identify whether licensed data is allowed in its evaluation set. During research university remediation, the proposed step to trace random sample records back to permitted originals and terms is not the direct restoration action for training data provenance when a team cannot identify whether licensed data is allowed in its evaluation set; the keyed repair is remove unsupported materials and replace them with authorized examples.

B:

During research university remediation, the proposed step to revoke excess access and invalidate cached sensitive retrieval results is not the direct restoration action for training data provenance when a team cannot identify whether licensed data is allowed in its evaluation set; the keyed repair is remove unsupported materials and replace them with authorized examples. If the team chose it instead, the targeted check to trace random sample records back to permitted originals and terms could still fail because the underlying required action is to remove unsupported materials and replace them with authorized examples.

C:

During research university remediation, the proposed step to assume every publicly reachable webpage is licensed for unrestricted reuse is not the direct restoration action for training data provenance when a team cannot identify whether licensed data is allowed in its evaluation set; the keyed repair is remove unsupported materials and replace them with authorized examples. After remediation the team must trace random sample records back to permitted originals and terms; the action in this option does not itself establish that target recovery outcome.

Question 244

A cloud developer at a education technology company wants the following issue to be prevented in later deployments: A team cannot identify whether licensed data is allowed in its evaluation set. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Assume every publicly reachable webpage is licensed for unrestricted reuse
  2. Remove unsupported materials and replace them with authorized examples
  3. Record source permission status and evaluation dataset revision history
  4. Gate promotion on safety regression criteria and alerting

Explanation

Correct Answer

C:

Standardizing the boundary to record source permission status and evaluation dataset revision history is preventive because the same training data provenance failure could otherwise return: a team cannot identify whether licensed data is allowed in its evaluation set. For this education technology company, the service should also capture dataset provenance and permitted use rights before ingestion, and tests should trace random sample records back to permitted originals and terms so that the ongoing requirement is measurable.

Incorrect Answers

A:

For prevention at the education technology company, the measure to assume every publicly reachable webpage is licensed for unrestricted reuse is weaker or aimed at another issue than the training data provenance guardrail to record source permission status and evaluation dataset revision history needed when a team cannot identify whether licensed data is allowed in its evaluation set. The lasting guardrail for this specific threat is to record source permission status and evaluation dataset revision history, not simply to complete an adjacent service task.

B:

The action to remove unsupported materials and replace them with authorized examples can contribute to training data provenance as a corrective remediation step; here, a team cannot identify whether licensed data is allowed in its evaluation set. For prevention at the education technology company, the measure to remove unsupported materials and replace them with authorized examples is weaker or aimed at another issue than the training data provenance guardrail to record source permission status and evaluation dataset revision history needed when a team cannot identify whether licensed data is allowed in its evaluation set.

D:

The action to gate promotion on safety regression criteria and alerting addresses safety drift regression rather than the decisive training data provenance condition. The education technology company should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to record source permission status and evaluation dataset revision history.

Question 245

During a reliability exercise, a site reliability engineer must accept a change for training data provenance. The issue to be addressed is: A team cannot identify whether licensed data is allowed in its evaluation set. Which concrete verification action demonstrates that the relevant behavior works?

  1. Inject source tables with deliberately absent totals and check for fabrication
  2. Assume every publicly reachable webpage is licensed for unrestricted reuse
  3. Trace random sample records back to permitted originals and terms
  4. Capture dataset provenance and permitted use rights before ingestion

Explanation

Correct Answer

C:

The targeted acceptance test is to trace random sample records back to permitted originals and terms. The design should first capture dataset provenance and permitted use rights before ingestion. A robust pharmaceutical research group deployment also needs to record source permission status and evaluation dataset revision history; passing this check provides relevant evidence but does not certify every possible input or security risk.

Incorrect Answers

A:

For hallucination evaluation, it can be sensible to inject source tables with deliberately absent totals and check for fabrication if a financial assistant invents currency figures absent from source statements. Without the actual change to trace random sample records back to permitted originals and terms, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

B:

In the pharmaceutical research group acceptance review, trying to assume every publicly reachable webpage is licensed for unrestricted reuse does not test the exact training data provenance condition where a team cannot identify whether licensed data is allowed in its evaluation set; the relevant test is to trace random sample records back to permitted originals and terms. Acceptance should instead demonstrate that the service can trace random sample records back to permitted originals and terms, using the described failure as the test criterion.

D:

The action to capture dataset provenance and permitted use rights before ingestion can contribute to training data provenance as a implementation step; here, a team cannot identify whether licensed data is allowed in its evaluation set. In the pharmaceutical research group acceptance review, trying to capture dataset provenance and permitted use rights before ingestion does not test the exact training data provenance condition where a team cannot identify whether licensed data is allowed in its evaluation set; the relevant test is to trace random sample records back to permitted originals and terms.

Question 246

At a software consultancy, the security engineer is preparing a service audit. The concrete requirement is: A financial assistant invents currency figures absent from source statements. Which design action BEST addresses the stated need?

  1. Require evidence-backed uncertainty and human approval before adverse decisions
  2. Use source-grounded evaluation to flag unsupported numeric assertions
  3. Judge factual accuracy only by fluent writing style and confidence
  4. Inject source tables with deliberately absent totals and check for fabrication

Explanation

Correct Answer

B:

Choosing to use source-grounded evaluation to flag unsupported numeric assertions directly targets the described hallucination evaluation gap: a financial assistant invents currency figures absent from source statements. As a lasting boundary, they also need to require evidence citation for material numerical claims, because an implementation by itself does not prove safe operation.

Incorrect Answers

A:

Choosing to require evidence-backed uncertainty and human approval before adverse decisions would direct effort toward false accusation risk, whose characteristic issue is that an incident triage agent accuses employees based on unverified telemetry. The relevant corrective or preventive action here remains to use source-grounded evaluation to flag unsupported numeric assertions, with its own validation criteria.

C:

Choosing to judge factual accuracy only by fluent writing style and confidence cannot directly replace the specified action to use source-grounded evaluation to flag unsupported numeric assertions when a financial assistant invents currency figures absent from source statements. The actual implementation must use source-grounded evaluation to flag unsupported numeric assertions, because the stated service condition is that a financial assistant invents currency figures absent from source statements.

D:

The action to inject source tables with deliberately absent totals and check for fabrication can contribute to hallucination evaluation as a targeted acceptance check step; here, a financial assistant invents currency figures absent from source statements. Choosing to inject source tables with deliberately absent totals and check for fabrication cannot directly replace the specified action to use source-grounded evaluation to flag unsupported numeric assertions when a financial assistant invents currency figures absent from source statements.

Question 247

After a security tabletop exercise at a regional hospital, the team observes this failure: A financial assistant invents currency figures absent from source statements. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Judge factual accuracy only by fluent writing style and confidence
  2. Use source-grounded evaluation to flag unsupported numeric assertions
  3. Inject source tables with deliberately absent totals and check for fabrication
  4. Treat a model's inferred user preference as permission to spend money

Explanation

Correct Answer

A:

The faulty assumption is judge factual accuracy only by fluent writing style and confidence. A functioning alternative would be to use source-grounded evaluation to flag unsupported numeric assertions, and the relevant validation would be to inject source tables with deliberately absent totals and check for fabrication.

Incorrect Answers

B:

The alternative to use source-grounded evaluation to flag unsupported numeric assertions is a implementation step rather than the faulty assumption being diagnosed. This regional hospital failure-analysis alternative of use source-grounded evaluation to flag unsupported numeric assertions does not diagnose the particular hallucination evaluation defect where a financial assistant invents currency figures absent from source statements; the keyed mistake is judge factual accuracy only by fluent writing style and confidence.

C:

The alternative to inject source tables with deliberately absent totals and check for fabrication is a targeted acceptance check step rather than the faulty assumption being diagnosed. This regional hospital failure-analysis alternative of inject source tables with deliberately absent totals and check for fabrication does not diagnose the particular hallucination evaluation defect where a financial assistant invents currency figures absent from source statements; the keyed mistake is judge factual accuracy only by fluent writing style and confidence.

D:

The option to treat a model's inferred user preference as permission to spend money would normally serve a different tool approval boundaries requirement, where an automated purchasing agent can place orders without user confirmation. The alternative may be useful in its own context, but it does not directly supply judge factual accuracy only by fluent writing style and confidence for the stated decision.

Question 248

A responsible-AI reviewer at a insurance platform is conducting a support escalation concerning hallucination evaluation. The observed problem is: A financial assistant invents currency figures absent from source statements. Which TWO actions together represent the needed implementation AND a durable preventive safeguard? (Select TWO answers.)

  1. Judge factual accuracy only by fluent writing style and confidence
  2. Inject source tables with deliberately absent totals and check for fabrication
  3. Use source-grounded evaluation to flag unsupported numeric assertions
  4. Require evidence citation for material numerical claims
  5. Escalate ambiguous high-impact decisions to a designated human reviewer

Explanation

Correct Answers

C:

Choosing to use source-grounded evaluation to flag unsupported numeric assertions directly targets the described hallucination evaluation gap: a financial assistant invents currency figures absent from source statements. As a lasting boundary, they also need to require evidence citation for material numerical claims, because an implementation by itself does not prove safe operation.

D:

Standardizing the boundary to require evidence citation for material numerical claims is preventive because the same hallucination evaluation failure could otherwise return: a financial assistant invents currency figures absent from source statements. For this insurance platform, the service should also use source-grounded evaluation to flag unsupported numeric assertions, and tests should inject source tables with deliberately absent totals and check for fabrication so that the ongoing requirement is measurable.

Incorrect Answers

A:

During insurance platform remediation, the proposed step to judge factual accuracy only by fluent writing style and confidence is not the direct restoration action for hallucination evaluation when a financial assistant invents currency figures absent from source statements; the keyed repair is use source-grounded evaluation to flag unsupported numeric assertions. After remediation the team must inject source tables with deliberately absent totals and check for fabrication; the action in this option does not itself establish that target recovery outcome.

B:

The action to inject source tables with deliberately absent totals and check for fabrication can contribute to hallucination evaluation as a targeted acceptance check step; here, a financial assistant invents currency figures absent from source statements. During insurance platform remediation, the proposed step to inject source tables with deliberately absent totals and check for fabrication is not the direct restoration action for hallucination evaluation when a financial assistant invents currency figures absent from source statements; the keyed repair is use source-grounded evaluation to flag unsupported numeric assertions.

E:

The action to escalate ambiguous high-impact decisions to a designated human reviewer addresses human escalation after uncertainty rather than the decisive hallucination evaluation condition. The insurance platform should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to use source-grounded evaluation to flag unsupported numeric assertions.

Question 249

During a service audit at a media platform, an AI team must address hallucination evaluation. The requirement is: A financial assistant invents currency figures absent from source statements. Which THREE actions together implement the needed solution, its preventive guardrail, and a targeted acceptance test? (Select THREE answers.)

  1. Use source-grounded evaluation to flag unsupported numeric assertions
  2. Inject source tables with deliberately absent totals and check for fabrication
  3. Require source-version traceability for compliance answers
  4. Require evidence citation for material numerical claims
  5. Repair retrieval and constrain output to verified facts or uncertainty
  6. Judge factual accuracy only by fluent writing style and confidence

Explanation

Correct Answers

A:

The proposed implementation is to use source-grounded evaluation to flag unsupported numeric assertions; this matches the operational requirement that a financial assistant invents currency figures absent from source statements. The team can inject source tables with deliberately absent totals and check for fabrication. The team should also require evidence citation for material numerical claims.

B:

The targeted acceptance test is to inject source tables with deliberately absent totals and check for fabrication. The design should first use source-grounded evaluation to flag unsupported numeric assertions. A robust media platform deployment also needs to require evidence citation for material numerical claims; passing this check provides relevant evidence but does not certify every possible input or security risk.

D:

The sustained prevention measure is to require evidence citation for material numerical claims. The implementation can still use the separate action to use source-grounded evaluation to flag unsupported numeric assertions, but a guardrail constrains future changes. The media platform can measure whether the rule works by arranging to inject source tables with deliberately absent totals and check for fabrication.

Incorrect Answers

C:

Choosing to require source-version traceability for compliance answers would direct effort toward source attribution controls, whose characteristic issue is that an agent quotes internal handbook statements without showing which version they came from. The relevant corrective or preventive action here remains to use source-grounded evaluation to flag unsupported numeric assertions, with its own validation criteria.

E:

The action to repair retrieval and constrain output to verified facts or uncertainty can contribute to hallucination evaluation as a corrective remediation step; here, a financial assistant invents currency figures absent from source statements. For prevention at the media platform, the measure to repair retrieval and constrain output to verified facts or uncertainty is weaker or aimed at another issue than the hallucination evaluation guardrail to use source-grounded evaluation to flag unsupported numeric assertions needed when a financial assistant invents currency figures absent from source statements.

F:

For prevention at the media platform, the measure to judge factual accuracy only by fluent writing style and confidence is weaker or aimed at another issue than the hallucination evaluation guardrail to use source-grounded evaluation to flag unsupported numeric assertions needed when a financial assistant invents currency figures absent from source statements. The lasting guardrail for this specific threat is to require evidence citation for material numerical claims, not simply to complete an adjacent service task.

Question 250

During a security tabletop exercise, a machine-learning developer must accept a change for hallucination evaluation. The issue to be addressed is: A financial assistant invents currency figures absent from source statements. Which concrete verification action demonstrates that the relevant behavior works?

  1. Judge factual accuracy only by fluent writing style and confidence
  2. Use source-grounded evaluation to flag unsupported numeric assertions
  3. Inject source tables with deliberately absent totals and check for fabrication
  4. Ask for a forbidden payroll change and inspect denial and audit evidence

Explanation

Correct Answer

C:

To verify this change, the customer support center should inject source tables with deliberately absent totals and check for fabrication. Operators still need the implementation to use source-grounded evaluation to flag unsupported numeric assertions and the preventive requirement to require evidence citation for material numerical claims.

Incorrect Answers

A:

In the customer support center acceptance review, trying to judge factual accuracy only by fluent writing style and confidence does not test the exact hallucination evaluation condition where a financial assistant invents currency figures absent from source statements; the relevant test is to inject source tables with deliberately absent totals and check for fabrication. Acceptance should instead demonstrate that the service can inject source tables with deliberately absent totals and check for fabrication, using the described failure as the test criterion.

B:

The action to use source-grounded evaluation to flag unsupported numeric assertions can contribute to hallucination evaluation as a implementation step; here, a financial assistant invents currency figures absent from source statements. In the customer support center acceptance review, trying to use source-grounded evaluation to flag unsupported numeric assertions does not test the exact hallucination evaluation condition where a financial assistant invents currency figures absent from source statements; the relevant test is to inject source tables with deliberately absent totals and check for fabrication.

D:

For agent scope escalation, it can be sensible to ask for a forbidden payroll change and inspect denial and audit evidence if a support agent uses a function tool to change unrelated payroll settings. Without the actual change to inject source tables with deliberately absent totals and check for fabrication, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

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