Microsoft AB-100 Dumps
Microsoft AB-100 Practice Test 67
Question 660
During a governance board meeting, an AI program manager must accept a change for retention minimization for agent logs. The issue to be addressed is: AI tool traces retain full chat histories indefinitely even though incidents need only minimal evidence. Which concrete verification action demonstrates that the relevant behavior works?
Explanation
Correct Answer
To verify this change, the municipal services department should inspect aged traces and confirm expired private payload fields are deleted. Operators still need the implementation to apply purpose-limited retention masking and least-privilege access to agent audit fields and the preventive requirement to set data-classified audit schemas and retention schedules before production.
Incorrect Answers
The action to set data-classified audit schemas and retention schedules before production can contribute to retention minimization for agent logs as a ongoing preventive rule step; here, aI tool traces retain full chat histories indefinitely even though incidents need only minimal evidence. The more directly responsive choice is to inspect aged traces and confirm expired private payload fields are deleted.
The assumption to store every sensitive prompt forever because richer logs might improve debugging is the exact failure pattern the team needs to avoid. The relevant change is to apply purpose-limited retention masking and least-privilege access to agent audit fields; a credible check must inspect aged traces and confirm expired private payload fields are deleted.
Choosing to challenge a rejection and verify the record links to actual criteria and evidence would direct effort toward ai accountability and explanation, whose characteristic issue is that an automated supplier rejection cannot identify the business policy or evidence that caused the result. The relevant corrective or preventive action here remains to inspect aged traces and confirm expired private payload fields are deleted, with its own validation criteria.
The action to apply purpose-limited retention masking and least-privilege access to agent audit fields can contribute to retention minimization for agent logs as a implementation step; here, aI tool traces retain full chat histories indefinitely even though incidents need only minimal evidence. The more directly responsive choice is to inspect aged traces and confirm expired private payload fields are deleted.
Question 661
At a regional hospital group, the security architect is preparing a data access design. The concrete requirement is: A customization pipeline accepts raw customer complaint text including credentials as tuning examples. Which design action BEST addresses the stated need?
Explanation
Correct Answer
The proposed implementation is to sanitize approve and restrict tuning data with lineage and appropriate rights; this matches the operational requirement that a customization pipeline accepts raw customer complaint text including credentials as tuning examples. The team can introduce a canary secret in a sample and confirm it cannot enter training. The team should also scan for secrets and record permitted training purposes before each run.
Incorrect Answers
Choosing to grant only the tool actions and record scope required for the business purpose would direct effort toward agent least-privilege tool scope, whose characteristic issue is that a service agent can read cases but also delete customer accounts through its broadly privileged tool. The relevant corrective or preventive action here remains to sanitize approve and restrict tuning data with lineage and appropriate rights, with its own validation criteria.
The assumption to fine-tune directly on every support transcript because model training is internal is the exact failure pattern the team needs to avoid. The relevant change is to sanitize approve and restrict tuning data with lineage and appropriate rights; a credible check must introduce a canary secret in a sample and confirm it cannot enter training.
The action to introduce a canary secret in a sample and confirm it cannot enter training can contribute to sensitive model tuning data as a targeted acceptance check step; here, a customization pipeline accepts raw customer complaint text including credentials as tuning examples. The more directly responsive choice is to sanitize approve and restrict tuning data with lineage and appropriate rights.
Question 662
After a customer case workflow redesign at an insurance provider, the team observes this failure: A customization pipeline accepts raw customer complaint text including credentials as tuning examples. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?
Explanation
Correct Answer
This design mistake is to fine-tune directly on every support transcript because model training is internal; it conflicts with the actual sensitive model tuning data behavior needed when a customization pipeline accepts raw customer complaint text including credentials as tuning examples. The insurance provider should instead sanitize approve and restrict tuning data with lineage and appropriate rights, followed by a check to introduce a canary secret in a sample and confirm it cannot enter training.
Incorrect Answers
The alternative to sanitize approve and restrict tuning data with lineage and appropriate rights is a implementation step rather than the faulty assumption being diagnosed. The failure in sensitive model tuning data arises from fine-tune directly on every support transcript because model training is internal; the concrete symptom is that a customization pipeline accepts raw customer complaint text including credentials as tuning examples.
The alternative to introduce a canary secret in a sample and confirm it cannot enter training is a targeted acceptance check step rather than the faulty assumption being diagnosed. The failure in sensitive model tuning data arises from fine-tune directly on every support transcript because model training is internal; the concrete symptom is that a customization pipeline accepts raw customer complaint text including credentials as tuning examples.
For the root-cause diagnosis decision on sensitive model tuning data with the option “Share one perpetual tenant administrator identity across all departmental agents”, the option to share one perpetual tenant administrator identity across all departmental agents would normally serve a different agent identity lifecycle requirement, where an unattended service principal continues using broad permissions after an agent project is discontinued. The alternative may be useful in its own context, but it does not directly supply fine-tune directly on every support transcript because model training is internal for the stated decision.
Question 663
The software publisher is responding to an AI adoption planning. A Dynamics 365 consultant discovers this issue: A customization pipeline accepts raw customer complaint text including credentials as tuning examples. What corrective step should the team take to restore the intended functionality most directly?
Explanation
Correct Answer
The direct repair is to quarantine the unauthorized examples and rebuild the dataset with controlled provenance. Once the condition is corrected, the software publisher should introduce a canary secret in a sample and confirm it cannot enter training. To reduce recurrence, it should also scan for secrets and record permitted training purposes before each run; those steps provide different recovery and prevention functions.
Incorrect Answers
The action to introduce a canary secret in a sample and confirm it cannot enter training can contribute to sensitive model tuning data as a targeted acceptance check step; here, a customization pipeline accepts raw customer complaint text including credentials as tuning examples. The more directly responsive choice is to quarantine the unauthorized examples and rebuild the dataset with controlled provenance.
The action to scan for secrets and record permitted training purposes before each run can contribute to sensitive model tuning data as a ongoing preventive rule step; here, a customization pipeline accepts raw customer complaint text including credentials as tuning examples. The more directly responsive choice is to quarantine the unauthorized examples and rebuild the dataset with controlled provenance.
For model artifact supply-chain trust, it can be sensible to remove the unreviewed model and redeploy a verified permitted artifact if a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. Without the actual change to quarantine the unauthorized examples and rebuild the dataset with controlled provenance, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.
The action to sanitize approve and restrict tuning data with lineage and appropriate rights can contribute to sensitive model tuning data as a implementation step; here, a customization pipeline accepts raw customer complaint text including credentials as tuning examples. The more directly responsive choice is to quarantine the unauthorized examples and rebuild the dataset with controlled provenance.
The assumption to fine-tune directly on every support transcript because model training is internal is the exact failure pattern the team needs to avoid. The relevant change is to sanitize approve and restrict tuning data with lineage and appropriate rights; a credible check must introduce a canary secret in a sample and confirm it cannot enter training.
Question 664
During a data access design at a government services agency, an AI solution team must address sensitive model tuning data. The requirement is: A customization pipeline accepts raw customer complaint text including credentials as tuning examples. Which THREE actions together implement the needed solution, its preventive guardrail, and a targeted acceptance test? (Select THREE answers.)
Explanation
Correct Answers
Standardizing the boundary to scan for secrets and record permitted training purposes before each run is preventive because the same sensitive model tuning data failure could otherwise return: a customization pipeline accepts raw customer complaint text including credentials as tuning examples. For this government services agency, the service should also sanitize approve and restrict tuning data with lineage and appropriate rights, and tests should introduce a canary secret in a sample and confirm it cannot enter training so that the ongoing requirement is measurable.
Choosing to sanitize approve and restrict tuning data with lineage and appropriate rights directly targets the described sensitive model tuning data gap: a customization pipeline accepts raw customer complaint text including credentials as tuning examples. As a lasting boundary, they also need to scan for secrets and record permitted training purposes before each run, because an implementation by itself does not prove safe operation.
To verify this change, the government services agency should introduce a canary secret in a sample and confirm it cannot enter training. Operators still need the implementation to sanitize approve and restrict tuning data with lineage and appropriate rights and the preventive requirement to scan for secrets and record permitted training purposes before each run.
Incorrect Answers
The action to quarantine the unauthorized examples and rebuild the dataset with controlled provenance can contribute to sensitive model tuning data as a corrective remediation step; here, a customization pipeline accepts raw customer complaint text including credentials as tuning examples. The more directly responsive choice is to sanitize approve and restrict tuning data with lineage and appropriate rights.
The assumption to fine-tune directly on every support transcript because model training is internal is the exact failure pattern the team needs to avoid. The relevant change is to sanitize approve and restrict tuning data with lineage and appropriate rights; a credible check must introduce a canary secret in a sample and confirm it cannot enter training.
Choosing to enforce least-privilege delegation and audit the full chain of authority would direct effort toward multi-agent delegated authority, whose characteristic issue is that a planning agent delegates tasks to an executor agent with more privileges than the original user. The relevant corrective or preventive action here remains to sanitize approve and restrict tuning data with lineage and appropriate rights, with its own validation criteria.
Question 665
During a customer case workflow redesign, a model evaluation specialist must accept a change for sensitive model tuning data. The issue to be addressed is: A customization pipeline accepts raw customer complaint text including credentials as tuning examples. Which concrete verification action demonstrates that the relevant behavior works?
Explanation
Correct Answer
The targeted acceptance test is to introduce a canary secret in a sample and confirm it cannot enter training. The design should first sanitize approve and restrict tuning data with lineage and appropriate rights. A robust manufacturing consortium deployment also needs to scan for secrets and record permitted training purposes before each run; passing this check provides relevant evidence but does not certify every possible input or security risk.
Incorrect Answers
For responsible ai harm assessment, it can be sensible to run difficult customer cases and examine harmful outcomes and proper human escalation if a customer-help agent may give discriminatory finance guidance to vulnerable customers. Without the actual change to introduce a canary secret in a sample and confirm it cannot enter training, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.
The assumption to fine-tune directly on every support transcript because model training is internal is the exact failure pattern the team needs to avoid. The relevant change is to sanitize approve and restrict tuning data with lineage and appropriate rights; a credible check must introduce a canary secret in a sample and confirm it cannot enter training.
The action to sanitize approve and restrict tuning data with lineage and appropriate rights can contribute to sensitive model tuning data as a implementation step; here, a customization pipeline accepts raw customer complaint text including credentials as tuning examples. The more directly responsive choice is to introduce a canary secret in a sample and confirm it cannot enter training.
Question 666
At a municipal services department, the product delivery lead is preparing a preproduction release gate. The concrete requirement is: An unattended service principal continues using broad permissions after an agent project is discontinued. Which design action BEST addresses the stated need?
Explanation
Correct Answer
Choosing to manage distinct workload identities with expiry review and revocation when agents retire directly targets the described agent identity lifecycle gap: an unattended service principal continues using broad permissions after an agent project is discontinued. As a lasting boundary, they also need to schedule owner attestations and automated retirement of agent identities, because an implementation by itself does not prove safe operation.
Incorrect Answers
The action to retire a test agent and verify its identity can no longer invoke production tools can contribute to agent identity lifecycle as a targeted acceptance check step; here, an unattended service principal continues using broad permissions after an agent project is discontinued. The more directly responsive choice is to manage distinct workload identities with expiry review and revocation when agents retire.
The assumption to share one perpetual tenant administrator identity across all departmental agents is the exact failure pattern the team needs to avoid. The relevant change is to manage distinct workload identities with expiry review and revocation when agents retire; a credible check must retire a test agent and verify its identity can no longer invoke production tools.
Choosing to document foreseeable harms impacted users mitigations and escalation routes before launch would direct effort toward responsible ai harm assessment, whose characteristic issue is that a customer-help agent may give discriminatory finance guidance to vulnerable customers. The relevant corrective or preventive action here remains to manage distinct workload identities with expiry review and revocation when agents retire, with its own validation criteria.
Question 667
After a customer service escalation review at a global distributor, the team observes this failure: An unattended service principal continues using broad permissions after an agent project is discontinued. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?
Explanation
Correct Answer
The faulty assumption is to share one perpetual tenant administrator identity across all departmental agents. A functioning alternative would be to manage distinct workload identities with expiry review and revocation when agents retire, and the relevant validation would be to retire a test agent and verify its identity can no longer invoke production tools.
Incorrect Answers
The alternative to manage distinct workload identities with expiry review and revocation when agents retire is a implementation step rather than the faulty assumption being diagnosed. The failure in agent identity lifecycle arises from share one perpetual tenant administrator identity across all departmental agents; the concrete symptom is that an unattended service principal continues using broad permissions after an agent project is discontinued.
The alternative to retire a test agent and verify its identity can no longer invoke production tools is a targeted acceptance check step rather than the faulty assumption being diagnosed. The failure in agent identity lifecycle arises from share one perpetual tenant administrator identity across all departmental agents; the concrete symptom is that an unattended service principal continues using broad permissions after an agent project is discontinued.
The option to treat delegation as unconditional permission escalation for downstream agents would normally serve a different multi-agent delegated authority requirement, where a planning agent delegates tasks to an executor agent with more privileges than the original user. The alternative may be useful in its own context, but it does not directly supply share one perpetual tenant administrator identity across all departmental agents for the stated decision.
Question 668
The digital payments provider is responding to an agent reliability assessment. An AI governance lead discovers this issue: An unattended service principal continues using broad permissions after an agent project is discontinued. What corrective step should the team take to restore the intended functionality most directly?
Explanation
Correct Answer
The appropriate remediation is to disable the stale principal and review its action history, because the failure concerns agent identity lifecycle: an unattended service principal continues using broad permissions after an agent project is discontinued. A meaningful verification should retire a test agent and verify its identity can no longer invoke production tools, and the durable operating rule should schedule owner attestations and automated retirement of agent identities after the immediate fix.
Incorrect Answers
The action to retire a test agent and verify its identity can no longer invoke production tools can contribute to agent identity lifecycle as a targeted acceptance check step; here, an unattended service principal continues using broad permissions after an agent project is discontinued. The more directly responsive choice is to disable the stale principal and review its action history.
This approach relates to sensitive model tuning data: a customization pipeline accepts raw customer complaint text including credentials as tuning examples. If the team chose it instead, the targeted check to retire a test agent and verify its identity can no longer invoke production tools could still fail because the underlying required action is to disable the stale principal and review its action history.
The assumption to share one perpetual tenant administrator identity across all departmental agents is the exact failure pattern the team needs to avoid. The relevant change is to manage distinct workload identities with expiry review and revocation when agents retire; a credible check must retire a test agent and verify its identity can no longer invoke production tools.
Question 669
A privacy engineer at a financial services firm wants the following issue to be prevented in later deployments: An unattended service principal continues using broad permissions after an agent project is discontinued. Which enduring control or engineering guardrail BEST prevents recurrence?
Explanation
Correct Answer
The sustained prevention measure is to schedule owner attestations and automated retirement of agent identities. The implementation can still use the separate action to manage distinct workload identities with expiry review and revocation when agents retire, but a guardrail constrains future changes. The financial services firm can measure whether the rule works by arranging to retire a test agent and verify its identity can no longer invoke production tools.
Incorrect Answers
The assumption to share one perpetual tenant administrator identity across all departmental agents is the exact failure pattern the team needs to avoid. The relevant change is to manage distinct workload identities with expiry review and revocation when agents retire; a credible check must retire a test agent and verify its identity can no longer invoke production tools.
The action to use destination allowlists DLP and redaction at connector boundaries addresses data exfiltration through tools rather than the decisive agent identity lifecycle condition. The financial services firm should not infer that fixing one adjacent business-system dependency validates its unrelated workflow; the requested action is to schedule owner attestations and automated retirement of agent identities.
The action to disable the stale principal and review its action history can contribute to agent identity lifecycle as a corrective remediation step; here, an unattended service principal continues using broad permissions after an agent project is discontinued. The more directly responsive choice is to schedule owner attestations and automated retirement of agent identities.
Question 670
During a customer service escalation review, a contact center architect must accept a change for agent identity lifecycle. The issue to be addressed is: An unattended service principal continues using broad permissions after an agent project is discontinued. Which concrete verification action demonstrates that the relevant behavior works?
Explanation
Correct Answer
To verify this change, the professional services partnership should retire a test agent and verify its identity can no longer invoke production tools. Operators still need the implementation to manage distinct workload identities with expiry review and revocation when agents retire and the preventive requirement to schedule owner attestations and automated retirement of agent identities.
Incorrect Answers
The assumption to share one perpetual tenant administrator identity across all departmental agents is the exact failure pattern the team needs to avoid. The relevant change is to manage distinct workload identities with expiry review and revocation when agents retire; a credible check must retire a test agent and verify its identity can no longer invoke production tools.
The action to manage distinct workload identities with expiry review and revocation when agents retire can contribute to agent identity lifecycle as a implementation step; here, an unattended service principal continues using broad permissions after an agent project is discontinued. The more directly responsive choice is to retire a test agent and verify its identity can no longer invoke production tools.
For model output data leakage, it can be sensible to probe known protected phrases and verify they are not disclosed to unauthorized users if an AI assistant accidentally includes confidential training examples in an answer. Without the actual change to retire a test agent and verify its identity can no longer invoke production tools, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.
Agentic AI Business Solutions Architect (AB-100) Practice Tests
- Microsoft AB-100 Practice Test 65 - 10 Questions (Last Reviewed 14-Nov-2026)
- Microsoft AB-100 Practice Test 66 - 10 Questions (Last Reviewed 13-Nov-2026)
- Microsoft AB-100 Practice Test 67 - 10 Questions (Last Reviewed 12-Nov-2026)
- Microsoft AB-100 Practice Test 68 - 10 Questions (Last Reviewed 11-Nov-2026)
- Microsoft AB-100 Practice Test 69 - 10 Questions (Last Reviewed 10-Nov-2026)