Practice Exams:

Microsoft AB-100 Dumps

EXAM: Agentic AI Business Solutions Architect
Verified By Experts
Premium Bundle
$49.99

AB-100 Premium ETE File


  • Premium File 129 Questions & Answers
Whats Included:
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates

Microsoft AB-100 Practice Test 64

Question 630

During a business AI architecture review, a Copilot Studio architect must accept a change for data exfiltration through tools. The issue to be addressed is: A customer-support agent can send unrestricted HTTP requests containing private case content. Which concrete verification action demonstrates that the relevant behavior works?

  1. Ask the agent to post a private case to a hostile URL and confirm denial
  2. Restrict outbound destinations payload fields and authorization of network tools
  3. Allow any user-provided URL to receive the full customer record for debugging
  4. Attempt an unapproved production publish and confirm tenant controls block it

Explanation

Correct Answer

A:

To verify this change, the insurance provider should ask the agent to post a private case to a hostile URL and confirm denial. Operators still need the implementation to restrict outbound destinations payload fields and authorization of network tools and the preventive requirement to use destination allowlists DLP and redaction at connector boundaries.

Incorrect Answers

B:

The action to restrict outbound destinations payload fields and authorization of network tools can contribute to data exfiltration through tools as a implementation step; here, a customer-support agent can send unrestricted HTTP requests containing private case content. The more directly responsive choice is to ask the agent to post a private case to a hostile URL and confirm denial.

C:

The assumption to allow any user-provided URL to receive the full customer record for debugging is the exact failure pattern the team needs to avoid. The relevant change is to restrict outbound destinations payload fields and authorization of network tools; a credible check must ask the agent to post a private case to a hostile URL and confirm denial.

D:

For governance of agent publication, it can be sensible to attempt an unapproved production publish and confirm tenant controls block it if department makers can publish experimental agents directly to the entire tenant without security review. Without the actual change to ask the agent to post a private case to a hostile URL and confirm denial, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

Question 631

At a retail franchise, the customer service architect is preparing a governance board meeting. The concrete requirement is: A user asks an AI assistant for commission data belonging to another sales representative. Which design action BEST addresses the stated need?

  1. Validate storage transit and processing geographies against binding data residency terms
  2. Rely solely on a prompt instruction telling the model not to show unrelated sales data
  3. Enforce row-level authorization in the retrieval and source service using requester identity
  4. Query two representatives' records and verify each sees only permitted rows

Explanation

Correct Answer

C:

The proposed implementation is to enforce row-level authorization in the retrieval and source service using requester identity; this matches the operational requirement that a user asks an AI assistant for commission data belonging to another sales representative. The team can query two representatives' records and verify each sees only permitted rows. The team should also validate effective data scopes for every search and tool request.

Incorrect Answers

A:

Choosing to validate storage transit and processing geographies against binding data residency terms would direct effort toward data residency and processing, whose characteristic issue is that an EU-controlled dataset is grounded through a service that might process prompts outside approved locations. The relevant corrective or preventive action here remains to enforce row-level authorization in the retrieval and source service using requester identity, with its own validation criteria.

B:

The assumption to rely solely on a prompt instruction telling the model not to show unrelated sales data is the exact failure pattern the team needs to avoid. The relevant change is to enforce row-level authorization in the retrieval and source service using requester identity; a credible check must query two representatives' records and verify each sees only permitted rows.

D:

The action to query two representatives' records and verify each sees only permitted rows can contribute to grounding-data row-level access as a targeted acceptance check step; here, a user asks an AI assistant for commission data belonging to another sales representative. The more directly responsive choice is to enforce row-level authorization in the retrieval and source service using requester identity.

Question 632

After a model evaluation review at a research institute, the team observes this failure: A user asks an AI assistant for commission data belonging to another sales representative. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Rely solely on a prompt instruction telling the model not to show unrelated sales data
  2. Share one perpetual tenant administrator identity across all departmental agents
  3. Query two representatives' records and verify each sees only permitted rows
  4. Enforce row-level authorization in the retrieval and source service using requester identity

Explanation

Correct Answer

A:

This design mistake is to rely solely on a prompt instruction telling the model not to show unrelated sales data; it conflicts with the actual grounding-data row-level access behavior needed when a user asks an AI assistant for commission data belonging to another sales representative. The research institute should instead enforce row-level authorization in the retrieval and source service using requester identity, followed by a check to query two representatives' records and verify each sees only permitted rows.

Incorrect Answers

B:

For the root-cause diagnosis decision on grounding-data row-level access 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 rely solely on a prompt instruction telling the model not to show unrelated sales data for the stated decision.

C:

The alternative to query two representatives' records and verify each sees only permitted rows is a targeted acceptance check step rather than the faulty assumption being diagnosed. The failure in grounding-data row-level access arises from rely solely on a prompt instruction telling the model not to show unrelated sales data; the concrete symptom is that a user asks an AI assistant for commission data belonging to another sales representative.

D:

The alternative to enforce row-level authorization in the retrieval and source service using requester identity is a implementation step rather than the faulty assumption being diagnosed. The failure in grounding-data row-level access arises from rely solely on a prompt instruction telling the model not to show unrelated sales data; the concrete symptom is that a user asks an AI assistant for commission data belonging to another sales representative.

Question 633

The hospitality operator is responding to an agent rollout. A security architect discovers this issue: A user asks an AI assistant for commission data belonging to another sales representative. What corrective step should the team take to restore the intended functionality most directly?

  1. Suspend the harmful workflow and apply evaluated corrective controls
  2. Correct overbroad retrieval identities and invalidate cached unauthorized results
  3. Query two representatives' records and verify each sees only permitted rows
  4. Rely solely on a prompt instruction telling the model not to show unrelated sales data

Explanation

Correct Answer

B:

The direct repair is to correct overbroad retrieval identities and invalidate cached unauthorized results. Once the condition is corrected, the hospitality operator should query two representatives' records and verify each sees only permitted rows. To reduce recurrence, it should also validate effective data scopes for every search and tool request; those steps provide different recovery and prevention functions.

Incorrect Answers

A:

This approach relates to responsible ai harm assessment: a customer-help agent may give discriminatory finance guidance to vulnerable customers. If the team chose it instead, the targeted check to query two representatives' records and verify each sees only permitted rows could still fail because the underlying required action is to correct overbroad retrieval identities and invalidate cached unauthorized results.

C:

The action to query two representatives' records and verify each sees only permitted rows can contribute to grounding-data row-level access as a targeted acceptance check step; here, a user asks an AI assistant for commission data belonging to another sales representative. The more directly responsive choice is to correct overbroad retrieval identities and invalidate cached unauthorized results.

D:

The assumption to rely solely on a prompt instruction telling the model not to show unrelated sales data is the exact failure pattern the team needs to avoid. The relevant change is to enforce row-level authorization in the retrieval and source service using requester identity; a credible check must query two representatives' records and verify each sees only permitted rows.

Question 634

An enterprise solution architect at a municipal services department wants the following issue to be prevented in later deployments: A user asks an AI assistant for commission data belonging to another sales representative. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Correct overbroad retrieval identities and invalidate cached unauthorized results
  2. Validate effective data scopes for every search and tool request
  3. Use destination allowlists DLP and redaction at connector boundaries
  4. Rely solely on a prompt instruction telling the model not to show unrelated sales data

Explanation

Correct Answer

B:

Standardizing the boundary to validate effective data scopes for every search and tool request is preventive because the same grounding-data row-level access failure could otherwise return: a user asks an AI assistant for commission data belonging to another sales representative. For this municipal services department, the service should also enforce row-level authorization in the retrieval and source service using requester identity, and tests should query two representatives' records and verify each sees only permitted rows so that the ongoing requirement is measurable.

Incorrect Answers

A:

The action to correct overbroad retrieval identities and invalidate cached unauthorized results can contribute to grounding-data row-level access as a corrective remediation step; here, a user asks an AI assistant for commission data belonging to another sales representative. The more directly responsive choice is to validate effective data scopes for every search and tool request.

C:

The action to use destination allowlists DLP and redaction at connector boundaries addresses data exfiltration through tools rather than the decisive grounding-data row-level access condition. The municipal services department should not infer that fixing one adjacent business-system dependency validates its unrelated workflow; the requested action is to validate effective data scopes for every search and tool request.

D:

The assumption to rely solely on a prompt instruction telling the model not to show unrelated sales data is the exact failure pattern the team needs to avoid. The relevant change is to enforce row-level authorization in the retrieval and source service using requester identity; a credible check must query two representatives' records and verify each sees only permitted rows.

Question 635

During a model evaluation review, a Dynamics 365 consultant must accept a change for grounding-data row-level access. The issue to be addressed is: A user asks an AI assistant for commission data belonging to another sales representative. Which concrete verification action demonstrates that the relevant behavior works?

  1. Enforce row-level authorization in the retrieval and source service using requester identity
  2. Tamper with a test model package and confirm promotion fails integrity checks
  3. Query two representatives' records and verify each sees only permitted rows
  4. Rely solely on a prompt instruction telling the model not to show unrelated sales data

Explanation

Correct Answer

C:

The targeted acceptance test is to query two representatives' records and verify each sees only permitted rows. The design should first enforce row-level authorization in the retrieval and source service using requester identity. A robust global distributor deployment also needs to validate effective data scopes for every search and tool request; passing this check provides relevant evidence but does not certify every possible input or security risk.

Incorrect Answers

A:

The action to enforce row-level authorization in the retrieval and source service using requester identity can contribute to grounding-data row-level access as a implementation step; here, a user asks an AI assistant for commission data belonging to another sales representative. The more directly responsive choice is to query two representatives' records and verify each sees only permitted rows.

B:

For model artifact supply-chain trust, it can be sensible to tamper with a test model package and confirm promotion fails integrity checks if a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. Without the actual change to query two representatives' records and verify each sees only permitted rows, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

D:

The assumption to rely solely on a prompt instruction telling the model not to show unrelated sales data is the exact failure pattern the team needs to avoid. The relevant change is to enforce row-level authorization in the retrieval and source service using requester identity; a credible check must query two representatives' records and verify each sees only permitted rows.

Question 636

At an insurance provider, the privacy engineer is preparing a customer case workflow redesign. The concrete requirement is: A team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. Which design action BEST addresses the stated need?

  1. Tamper with a test model package and confirm promotion fails integrity checks
  2. Assume a popular model download is automatically trusted for regulated operations
  3. Assess model provenance license integrity and risk before any sensitive deployment
  4. Assess accessibility and language support against affected user groups before production

Explanation

Correct Answer

C:

Choosing to assess model provenance license integrity and risk before any sensitive deployment directly targets the described model artifact supply-chain trust gap: a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. As a lasting boundary, they also need to require signed artifact origin scanning and review of model dependencies, because an implementation by itself does not prove safe operation.

Incorrect Answers

A:

The action to tamper with a test model package and confirm promotion fails integrity checks can contribute to model artifact supply-chain trust as a targeted acceptance check step; here, a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. The more directly responsive choice is to assess model provenance license integrity and risk before any sensitive deployment.

B:

The assumption to assume a popular model download is automatically trusted for regulated operations is the exact failure pattern the team needs to avoid. The relevant change is to assess model provenance license integrity and risk before any sensitive deployment; a credible check must tamper with a test model package and confirm promotion fails integrity checks.

D:

Choosing to assess accessibility and language support against affected user groups before production would direct effort toward responsible ai review of accessibility, whose characteristic issue is that an AI-powered customer portal works for typed English input but fails users relying on accessibility features. The relevant corrective or preventive action here remains to assess model provenance license integrity and risk before any sensitive deployment, with its own validation criteria.

Question 637

After an AI adoption planning at a software publisher, the team observes this failure: A team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Declare every AI risk permanently accepted after an informal project meeting
  2. Assess model provenance license integrity and risk before any sensitive deployment
  3. Assume a popular model download is automatically trusted for regulated operations
  4. Tamper with a test model package and confirm promotion fails integrity checks

Explanation

Correct Answer

C:

The faulty assumption is to assume a popular model download is automatically trusted for regulated operations. A functioning alternative would be to assess model provenance license integrity and risk before any sensitive deployment, and the relevant validation would be to tamper with a test model package and confirm promotion fails integrity checks.

Incorrect Answers

A:

The option to declare every AI risk permanently accepted after an informal project meeting would normally serve a different agent risk exception governance requirement, where a business owner accepts residual risk for an agent without specifying duration or compensating measures. The alternative may be useful in its own context, but it does not directly supply assume a popular model download is automatically trusted for regulated operations for the stated decision.

B:

The alternative to assess model provenance license integrity and risk before any sensitive deployment is a implementation step rather than the faulty assumption being diagnosed. The failure in model artifact supply-chain trust arises from assume a popular model download is automatically trusted for regulated operations; the concrete symptom is that a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data.

D:

The alternative to tamper with a test model package and confirm promotion fails integrity checks is a targeted acceptance check step rather than the faulty assumption being diagnosed. The failure in model artifact supply-chain trust arises from assume a popular model download is automatically trusted for regulated operations; the concrete symptom is that a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data.

Question 638

The government services agency is responding to a data access design. A product delivery lead discovers this issue: A team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. What corrective step should the team take to restore the intended functionality most directly?

  1. Remove the unreviewed model and redeploy a verified permitted artifact
  2. Assume a popular model download is automatically trusted for regulated operations
  3. Tamper with a test model package and confirm promotion fails integrity checks
  4. Require signed artifact origin scanning and review of model dependencies
  5. Revoke the delete grant and audit past administrative calls

Explanation

Correct Answer

A:

The appropriate remediation is to remove the unreviewed model and redeploy a verified permitted artifact, because the failure concerns model artifact supply-chain trust: a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. A meaningful verification should tamper with a test model package and confirm promotion fails integrity checks, and the durable operating rule should require signed artifact origin scanning and review of model dependencies after the immediate fix.

Incorrect Answers

B:

The assumption to assume a popular model download is automatically trusted for regulated operations is the exact failure pattern the team needs to avoid. The relevant change is to assess model provenance license integrity and risk before any sensitive deployment; a credible check must tamper with a test model package and confirm promotion fails integrity checks.

C:

The action to tamper with a test model package and confirm promotion fails integrity checks can contribute to model artifact supply-chain trust as a targeted acceptance check step; here, a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. The more directly responsive choice is to remove the unreviewed model and redeploy a verified permitted artifact.

D:

The action to require signed artifact origin scanning and review of model dependencies can contribute to model artifact supply-chain trust as a ongoing preventive rule step; here, a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. The more directly responsive choice is to remove the unreviewed model and redeploy a verified permitted artifact.

E:

The action to revoke the delete grant and audit past administrative calls addresses agent least-privilege tool scope rather than the decisive model artifact supply-chain trust condition. The government services agency should not infer that fixing one adjacent business-system dependency validates its unrelated workflow; the requested action is to remove the unreviewed model and redeploy a verified permitted artifact.

Question 639

A business application owner at a manufacturing consortium wants the following issue to be prevented in later deployments: A team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Require signed artifact origin scanning and review of model dependencies
  2. Track exception expiry conditions and accountable approval roles
  3. Assume a popular model download is automatically trusted for regulated operations
  4. Remove the unreviewed model and redeploy a verified permitted artifact

Explanation

Correct Answer

A:

The sustained prevention measure is to require signed artifact origin scanning and review of model dependencies. The implementation can still use the separate action to assess model provenance license integrity and risk before any sensitive deployment, but a guardrail constrains future changes. The manufacturing consortium can measure whether the rule works by arranging to tamper with a test model package and confirm promotion fails integrity checks.

Incorrect Answers

B:

The action to track exception expiry conditions and accountable approval roles addresses agent risk exception governance rather than the decisive model artifact supply-chain trust condition. The manufacturing consortium should not infer that fixing one adjacent business-system dependency validates its unrelated workflow; the requested action is to require signed artifact origin scanning and review of model dependencies.

C:

The assumption to assume a popular model download is automatically trusted for regulated operations is the exact failure pattern the team needs to avoid. The relevant change is to assess model provenance license integrity and risk before any sensitive deployment; a credible check must tamper with a test model package and confirm promotion fails integrity checks.

D:

The action to remove the unreviewed model and redeploy a verified permitted artifact can contribute to model artifact supply-chain trust as a corrective remediation step; here, a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. The more directly responsive choice is to require signed artifact origin scanning and review of model dependencies.

Question 640

During an AI adoption planning, an AI governance lead must accept a change for model artifact supply-chain trust. The issue to be addressed is: A team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. Which concrete verification action demonstrates that the relevant behavior works?

  1. Assess model provenance license integrity and risk before any sensitive deployment
  2. Probe known protected phrases and verify they are not disclosed to unauthorized users
  3. Assume a popular model download is automatically trusted for regulated operations
  4. Tamper with a test model package and confirm promotion fails integrity checks

Explanation

Correct Answer

D:

To verify this change, the transport services operator should tamper with a test model package and confirm promotion fails integrity checks. Operators still need the implementation to assess model provenance license integrity and risk before any sensitive deployment and the preventive requirement to require signed artifact origin scanning and review of model dependencies.

Incorrect Answers

A:

The action to assess model provenance license integrity and risk before any sensitive deployment can contribute to model artifact supply-chain trust as a implementation step; here, a team downloads a community model with unknown origin and attaches it to sensitive Dynamics data. The more directly responsive choice is to tamper with a test model package and confirm promotion fails integrity checks.

B:

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 tamper with a test model package and confirm promotion fails integrity checks, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

C:

The assumption to assume a popular model download is automatically trusted for regulated operations is the exact failure pattern the team needs to avoid. The relevant change is to assess model provenance license integrity and risk before any sensitive deployment; a credible check must tamper with a test model package and confirm promotion fails integrity checks.

Agentic AI Business Solutions Architect (AB-100) Practice Tests