Microsoft AB-100 Exam Dumps, Practice Test Questions

100% Latest & Updated Microsoft AB-100 Practice Test Questions, Exam Dumps & Verified Answers!
30 Days Free Updates, Instant Download!

Microsoft AB-100  Premium File
$54.99
$49.99

AB-100 Premium File

  • Premium File: 129 Questions & Answers. Last update: Nov 21, 2026
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates

AB-100 Premium File

Microsoft AB-100  Premium File
  • Premium File: 129 Questions & Answers. Last update: Nov 21, 2026
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates
$54.99
$49.99

Microsoft AB-100 Practice Test 37

Question 360

An agent accepts a forged user role from the text of a message without checking the caller identity. Which solution approach best meets the stated requirement?

  1. Derive authorization from validated identity claims, never from prompt content.
  2. Use delegated user authentication for content protected by individual permissions.
  3. Enforce caller-scoped row and regional permissions in retrieval.
  4. Use a narrowly scoped service identity appropriate for the low-risk lookup.

Explanation

Correct Answer

A:

Natural-language role claims are untrusted and cannot expand tool privileges.

Incorrect Answers

B:

User-bound identity stops the agent from inheriting inappropriate maker access. Natural-language role claims are untrusted and cannot expand tool privileges.

C:

Source authorization must constrain what the specific user can access, but natural-language role claims are untrusted and cannot expand tool privileges.

D:

A bounded service credential can fit a shared public lookup when it grants no sensitive rights. Natural-language role claims are untrusted and cannot expand tool privileges.

Question 361

A product page retrieved for price lookup includes hidden instructions to send the customer’s address to a third-party URL. Which design choice targets the observed problem?

  1. Validate tool outputs as data against the declared schema and ignore embedded control directives.
  2. Review model provenance, license and integrity before approving deployment.
  3. Treat retrieved text as untrusted and enforce outbound destination restrictions.
  4. Keep secrets and privileged instructions isolated from response generation.

Explanation

Correct Answer

C:

Document text must not become a privileged agent instruction or exfiltration command.

Incorrect Answers

A:

Tool output has lower trust than the configured execution policy. Document text must not become a privileged agent instruction or exfiltration command.

B:

Third-party model supply-chain checks reduce security and legal uncertainty, but document text must not become a privileged agent instruction or exfiltration command.

D:

Sensitive runtime information is not authorized user content; document text must not become a privileged agent instruction or exfiltration command.

Question 362

A user asks a legal assistant to reveal its system prompt and protected service account credentials. Which concrete solution decision should the team make?

  1. Allowlist reachable destinations and validate URLs before tool execution.
  2. Use delegated user authentication for content protected by individual permissions.
  3. Validate tool outputs as data against the declared schema and ignore embedded control directives.
  4. Keep secrets and privileged instructions isolated from response generation.

Explanation

Correct Answer

D:

Sensitive runtime information is not authorized user content.

Incorrect Answers

A:

Network and URL restrictions limit server-side request forgery and data exfiltration, but sensitive runtime information is not authorized user content.

B:

User-bound identity stops the agent from inheriting inappropriate maker access. Sensitive runtime information is not authorized user content.

C:

Tool output has lower trust than the configured execution policy; sensitive runtime information is not authorized user content.

Question 363

A tool response contains YAML that pretends to redefine the tool schema and bypass a required approval. Which design decision should the architect prioritize?

  1. Validate tool outputs as data against the declared schema and ignore embedded control directives.
  2. Evaluate retrieval poisoning risk and apply source trust/approval controls.
  3. Treat retrieved text as untrusted and enforce outbound destination restrictions.
  4. Allowlist reachable destinations and validate URLs before tool execution.

Explanation

Correct Answer

A:

Tool output has lower trust than the configured execution policy.

Incorrect Answers

B:

External content can be altered to influence agent answers and tool choices; tool output has lower trust than the configured execution policy.

C:

Document text must not become a privileged agent instruction or exfiltration command, whereas tool output has lower trust than the configured execution policy.

D:

Network and URL restrictions limit server-side request forgery and data exfiltration. Tool output has lower trust than the configured execution policy.

Question 364

A business agent autonomously follows any URL returned by a model, risking requests to private network endpoints. Which controlled change is most appropriate for this situation?

  1. Treat retrieved text as untrusted and enforce outbound destination restrictions.
  2. Allowlist reachable destinations and validate URLs before tool execution.
  3. Keep secrets and privileged instructions isolated from response generation.
  4. Expose an allowlisted minimal tool catalog per agent role.

Explanation

Correct Answer

B:

Network and URL restrictions limit server-side request forgery and data exfiltration.

Incorrect Answers

A:

Document text must not become a privileged agent instruction or exfiltration command; network and URL restrictions limit server-side request forgery and data exfiltration.

C:

Sensitive runtime information is not authorized user content. Network and URL restrictions limit server-side request forgery and data exfiltration.

D:

Tool discovery itself should not grant irrelevant sensitive capabilities, but network and URL restrictions limit server-side request forgery and data exfiltration.

Question 365

A healthcare solution stores records in an approved region but sends model prompts to an unapproved processing region. What is the most defensible intervention for this case?

  1. Apply location and minimization controls to telemetry and audit storage.
  2. Revoke exposed credentials and use secure secret management with minimum scoped access.
  3. Assess and constrain data flows across all integrated external services.
  4. Review processing and inference locations, not only stored data location.

Explanation

Correct Answer

D:

Residency applies to data in transit and processing as well as persistent storage.

Incorrect Answers

A:

Observability data may contain protected information and require the same governance review, whereas residency applies to data in transit and processing as well as persistent storage.

B:

Credential exposure can enable unauthorized inference and data access. Residency applies to data in transit and processing as well as persistent storage.

C:

Cross-service tool calls can cause regulated transfers even when core hosting is compliant, but residency applies to data in transit and processing as well as persistent storage.

Question 366

Audit logs copy full patient prompts to a global monitoring service without a residency assessment. Which architecture response most directly addresses the risk?

  1. Apply location and minimization controls to telemetry and audit storage.
  2. Evaluate fairness disparities and require accountable review of adverse outcomes.
  3. Assess and constrain data flows across all integrated external services.
  4. Partition sources and execution routes according to data classification and region.

Explanation

Correct Answer

A:

Observability data may contain protected information and require the same governance review.

Incorrect Answers

B:

Group-level error analysis can reveal differential impact hidden by aggregate accuracy, but observability data may contain protected information and require the same governance review.

C:

Cross-service tool calls can cause regulated transfers even when core hosting is compliant. Observability data may contain protected information and require the same governance review.

D:

Classification-aware routing avoids applying the weakest residency rule to protected content; observability data may contain protected information and require the same governance review.

Question 367

An agent reads customer records in Europe but uses a third-party tool that transfers them abroad. Which recommendation addresses the identified failure most directly?

  1. Review processing and inference locations, not only stored data location.
  2. Assess and constrain data flows across all integrated external services.
  3. Expose an allowlisted minimal tool catalog per agent role.
  4. Partition sources and execution routes according to data classification and region.

Explanation

Correct Answer

B:

Cross-service tool calls can cause regulated transfers even when core hosting is compliant.

Incorrect Answers

A:

Residency applies to data in transit and processing as well as persistent storage, but cross-service tool calls can cause regulated transfers even when core hosting is compliant.

C:

Tool discovery itself should not grant irrelevant sensitive capabilities; cross-service tool calls can cause regulated transfers even when core hosting is compliant.

D:

Classification-aware routing avoids applying the weakest residency rule to protected content, whereas cross-service tool calls can cause regulated transfers even when core hosting is compliant.

Question 368

A company has different residency restrictions for regulated and nonregulated knowledge collections. What should the solution architect recommend for this specific need?

  1. Allowlist reachable destinations and validate URLs before tool execution.
  2. Review processing and inference locations, not only stored data location.
  3. Partition sources and execution routes according to data classification and region.
  4. Apply location and minimization controls to telemetry and audit storage.

Explanation

Correct Answer

C:

Classification-aware routing avoids applying the weakest residency rule to protected content.

Incorrect Answers

A:

Network and URL restrictions limit server-side request forgery and data exfiltration. Classification-aware routing avoids applying the weakest residency rule to protected content.

B:

Residency applies to data in transit and processing as well as persistent storage; classification-aware routing avoids applying the weakest residency rule to protected content.

D:

Observability data may contain protected information and require the same governance review. Classification-aware routing avoids applying the weakest residency rule to protected content.

Question 369

A knowledge index contains documents from several departments and the agent assumes index membership grants universal access. Which approach best satisfies the actual business constraint?

  1. Enforce document-level access control during retrieval for every caller.
  2. Exclude or minimize protected data until lawful purpose and access authorization are established.
  3. Deidentify test and evaluation datasets and scope access to reviewers.
  4. Capture authorized retrieval events with subject, source and purpose metadata.

Explanation

Correct Answer

A:

Indexing does not erase the underlying source permissions.

Incorrect Answers

B:

Fine-tuning use needs its own permitted basis; access for operations is not automatic consent for training, whereas indexing does not erase the underlying source permissions.

C:

Evaluation does not require unrestricted production PII exposure. Indexing does not erase the underlying source permissions.

D:

Access auditing can demonstrate when protected information was used by the assistant, but indexing does not erase the underlying source permissions.

Question 370

A model-tuning experiment includes raw employee salaries without an approved data use purpose. Which specific remediation should the business adopt?

  1. Deidentify test and evaluation datasets and scope access to reviewers.
  2. Exclude or minimize protected data until lawful purpose and access authorization are established.
  3. Validate data lineage and inspect anomalous training inputs before model promotion.
  4. Preserve source classification and apply output handling controls to retrieved fragments.

Explanation

Correct Answer

B:

Fine-tuning use needs its own permitted basis; access for operations is not automatic consent for training.

Incorrect Answers

A:

Evaluation does not require unrestricted production PII exposure, whereas fine-tuning use needs its own permitted basis; access for operations is not automatic consent for training.

C:

Poisoning defense requires trust in the data pipeline, not just final accuracy tests, but fine-tuning use needs its own permitted basis; access for operations is not automatic consent for training.

D:

Security labels must follow sensitive content through the generated response path. Fine-tuning use needs its own permitted basis; access for operations is not automatic consent for training.

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

UP

SPECIAL OFFER: GET 10% OFF

This is ONE TIME OFFER

ExamSnap Discount Offer
Enter Your Email Address to Receive Your 10% Off Discount Code

A confirmation link will be sent to this email address to verify your login. *We value your privacy. We will not rent or sell your email address.

Download Free Demo of VCE Exam Simulator

Experience Avanset VCE Exam Simulator for yourself.

Simply submit your e-mail address below to get started with our interactive software demo of your free trial.

Free Demo Limits: In the demo version you will be able to access only first 5 questions from exam.