Practice Exams:

IAPP AIGP Dumps

EXAM: Artificial Intelligence Governance Professional
Verified By Experts
Premium Bundle
$40.00
$59.97 $99.97

AIGP Premium Bundle


  • Premium File 222 Questions & Answers
  • Training Course 124 Lectures
  • Study Guide 566 Pages

IAPP AIGP Practice Test 20

Question 190

The governance team needs both a decision and a way to verify it for this case: A nonprofit plans to use donor contact records collected for receipts to score volunteers for eligibility. Which pair is best?

  1. Action: Design a process to identify relevant data subject requests and assess applicable automated-decision safeguards. Evidence: Test sample requests for access, correction and review against the governing jurisdictional rules and operational logs.
  2. Action: Test whether a proposed new model purpose is compatible with the reason personal information was originally obtained. Evidence: Record the original notice and collection purpose alongside the proposed use and compatibility determination.
  3. Action: Escalate biometric or other specially protected personal data for a jurisdiction-specific legal and security review. Evidence: Confirm the applicable category, explicit necessity assessment, protective controls and approved processing conditions.
  4. Action: Check the provider's contract and relevant law for permitted commercial use and responsibility for generated content. Evidence: Review license terms and approval records for representative public-facing generated materials.

Explanation

Correct Answer

B:

For the action and verification, test whether a proposed new model purpose is compatible with the reason personal information was originally obtained. Record the original notice and collection purpose alongside the proposed use and compatibility determination. Together, they implement and test the intended safeguard. Reusing information for an unrelated objective can create a new privacy obligation despite good technical safeguards

Incorrect Answers

A:

For the action and verification, the alternative—design a process to identify relevant data subject requests and assess applicable automated-decision safeguards. Evidence: Test sample requests for access, correction and review against the governing jurisdictional rules and operational logs—would not resolve the stated issue. Reusing information for an unrelated objective can create a new privacy obligation despite good technical safeguards

C:

The alternative—escalate biometric or other specially protected personal data for a jurisdiction-specific legal and security review. Evidence: Confirm the applicable category, explicit necessity assessment, protective controls and approved processing conditions—would not resolve the stated issue. Reusing information for an unrelated objective can create a new privacy obligation despite good technical safeguards.

D:

The alternative—check the provider's contract and relevant law for permitted commercial use and responsibility for generated content. Evidence: Review license terms and approval records for representative public-facing generated materials—would not resolve the stated issue. Reusing information for an unrelated objective can create a new privacy obligation despite good technical safeguards.

Question 191

A hotel is developing a service-demand forecast using bookings but considers ingesting passport numbers. Which step would provide the most targeted control for the problem described?

  1. Test whether a proposed new model purpose is compatible with the reason personal information was originally obtained.
  2. Identify and document the applicable lawful basis before reusing personal information for model development.
  3. Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient.
  4. Conduct a jurisdiction- and use-specific prohibited-practice screen before selecting a use case for development.

Explanation

Correct Answer

C:

Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. Collecting more personal data than required increases exposure without automatically improving the model. The practical acceptance evidence is: compare approved feature requirements with actual datasets and test a reduced-data version for utility.

Incorrect Answers

A:

This would help address a separate governance task: Test whether a proposed new model purpose is compatible with the reason personal information was originally obtained. For this case, however, collecting more personal data than required increases exposure without automatically improving the model. The more direct step is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient.

B:

This would help address a separate governance task: Identify and document the applicable lawful basis before reusing personal information for model development. For this case, however, collecting more personal data than required increases exposure without automatically improving the model. The more direct step is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient.

D:

This would help address a separate governance task: Conduct a jurisdiction- and use-specific prohibited-practice screen before selecting a use case for development. For this case, however, collecting more personal data than required increases exposure without automatically improving the model. The more direct step is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient.

Question 192

The organization chooses to remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient for this situation: A hotel is developing a service-demand forecast using bookings but considers ingesting passport numbers. Which rationale best supports that choice?

  1. Sensitive data can trigger heightened restrictions and requires particular attention to necessity, safeguards and permitted uses.
  2. Reusing information for an unrelated objective can create a new privacy obligation despite good technical safeguards.
  3. These functions provide complementary steps: establish context, assess relevant risks and respond to the findings.
  4. Collecting more personal data than required increases exposure without automatically improving the model.

Explanation

Correct Answer

D:

Collecting more personal data than required increases exposure without automatically improving the model. That is why the case calls for this action: remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient.

Incorrect Answers

A:

That explains why a team might escalate biometric or other specially protected personal data for a jurisdiction-specific legal and security review, but it does not justify the decision required by the case. Collecting more personal data than required increases exposure without automatically improving the model.

B:

That explains why a team might test whether a proposed new model purpose is compatible with the reason personal information was originally obtained, but it does not justify the decision required by the case. Collecting more personal data than required increases exposure without automatically improving the model.

C:

That explains why a team might map AI context, measure identified risks and manage prioritized responses using the NIST AI RMF functions, but it does not justify the decision required by the case. Collecting more personal data than required increases exposure without automatically improving the model.

Question 193

A hotel is developing a service-demand forecast using bookings but considers ingesting passport numbers. After taking the recommended governance action, which evidence most directly verifies that the risk has been addressed?

  1. Compare approved feature requirements with actual datasets and test a reduced-data version for utility.
  2. Check the data-use assessment for jurisdiction, purpose, applicable lawful basis and approval before ingestion.
  3. Confirm the applicable category, explicit necessity assessment, protective controls and approved processing conditions.
  4. Check the legal classification and an obligation-by-obligation evidence matrix before release.

Explanation

Correct Answer

A:

Compare approved feature requirements with actual datasets and test a reduced-data version for utility. This verifies the necessary safeguard rather than merely showing that an AI project was completed. Collecting more personal data than required increases exposure without automatically improving the model.

Incorrect Answers

B:

Check the data-use assessment for jurisdiction, purpose, applicable lawful basis and approval before ingestion would test a different outcome. The directly relevant evidence is to compare approved feature requirements with actual datasets and test a reduced-data version for utility.

C:

Confirm the applicable category, explicit necessity assessment, protective controls and approved processing conditions would test a different outcome. The directly relevant evidence is to compare approved feature requirements with actual datasets and test a reduced-data version for utility.

D:

Check the legal classification and an obligation-by-obligation evidence matrix before release would test a different outcome. The directly relevant evidence is to compare approved feature requirements with actual datasets and test a reduced-data version for utility.

Question 194

The organization has this problem: A hotel is developing a service-demand forecast using bookings but considers ingesting passport numbers. It takes a shortcut: Feed the entire customer database into the model because storage and compute are available. What is the principal risk?

  1. Identifiable or sensitive characteristics may still be exposed or processed unlawfully.
  2. Unneeded sensitive fields can leak or influence outcomes while adding no justified benefit.
  3. An unauthorized new processing purpose can be hidden behind existing access rights.
  4. The organization may be unable to show compliance or respond effectively to a formal inquiry.

Explanation

Correct Answer

B:

Unneeded sensitive fields can leak or influence outcomes while adding no justified benefit. The shortcut overlooks the central safeguard: remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. Collecting more personal data than required increases exposure without automatically improving the model.

Incorrect Answers

A:

This consequence follows more directly from another shortcut: Treat face embeddings as ordinary anonymous telemetry simply because names are removed. The shortcut actually described would instead create this risk: Unneeded sensitive fields can leak or influence outcomes while adding no justified benefit.

C:

This consequence follows more directly from another shortcut: Assume a dataset may be trained on because staff already can open the underlying system. The shortcut actually described would instead create this risk: Unneeded sensitive fields can leak or influence outcomes while adding no justified benefit.

D:

This consequence follows more directly from another shortcut: Wait for a regulator to ask questions before creating system documentation and records. The shortcut actually described would instead create this risk: Unneeded sensitive fields can leak or influence outcomes while adding no justified benefit.

Question 195

A hotel is developing a service-demand forecast using bookings but considers ingesting passport numbers. Which action and evidence pair most directly addresses the situation?

  1. Action: Test whether a proposed new model purpose is compatible with the reason personal information was originally obtained. Evidence: Record the original notice and collection purpose alongside the proposed use and compatibility determination.
  2. Action: Identify and document the applicable lawful basis before reusing personal information for model development. Evidence: Check the data-use assessment for jurisdiction, purpose, applicable lawful basis and approval before ingestion.
  3. Action: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. Evidence: Compare approved feature requirements with actual datasets and test a reduced-data version for utility.
  4. Action: Differentiate provider, deployer, importer and distributor responsibilities for the AI supply arrangement under applicable law. Evidence: Map contracts, modifications, branding and deployment acts to the role definitions used in the relevant regime.

Explanation

Correct Answer

C:

Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. Compare approved feature requirements with actual datasets and test a reduced-data version for utility. Together, they implement and test the intended safeguard. Collecting more personal data than required increases exposure without automatically improving the model.

Incorrect Answers

A:

This pair supports another governance intervention: Test whether a proposed new model purpose is compatible with the reason personal information was originally obtained. For the actual case, the required action is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. The supporting evidence is: Compare approved feature requirements with actual datasets and test a reduced-data version for utility.

B:

This pair supports another governance intervention: Identify and document the applicable lawful basis before reusing personal information for model development. For the actual case, the required action is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. The supporting evidence is: Compare approved feature requirements with actual datasets and test a reduced-data version for utility.

D:

This pair supports another governance intervention: Differentiate provider, deployer, importer and distributor responsibilities for the AI supply arrangement under applicable law. For the actual case, the required action is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. The supporting evidence is: Compare approved feature requirements with actual datasets and test a reduced-data version for utility.

Question 196

An education provider wants a help-desk classifier and proposes adding pupils' home addresses. Which governance action most directly addresses the need in this situation?

  1. Escalate biometric or other specially protected personal data for a jurisdiction-specific legal and security review.
  2. Design a process to identify relevant data subject requests and assess applicable automated-decision safeguards.
  3. Identify whether an AI interaction or generated content requires specific disclosure in the applicable context.
  4. Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient.

Explanation

Correct Answer

D:

For the decision, remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. Collecting more personal data than required increases exposure without automatically improving the model. The practical acceptance evidence is: compare approved feature requirements with actual datasets and test a reduced-data version for utility

Incorrect Answers

A:

This would help address a separate governance task: Escalate biometric or other specially protected personal data for a jurisdiction-specific legal and security review. For this case, however, collecting more personal data than required increases exposure without automatically improving the model. The more direct step is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient.

B:

This would help address a separate governance task: Design a process to identify relevant data subject requests and assess applicable automated-decision safeguards. For this case, however, collecting more personal data than required increases exposure without automatically improving the model. The more direct step is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient.

C:

This would help address a separate governance task: Identify whether an AI interaction or generated content requires specific disclosure in the applicable context. For this case, however, collecting more personal data than required increases exposure without automatically improving the model. The more direct step is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient.

Question 197

A team reviewing the following situation asks for the most relevant governance principle: An education provider wants a help-desk classifier and proposes adding pupils' home addresses. Which principle best explains the required control?

  1. Collecting more personal data than required increases exposure without automatically improving the model.
  2. AI-supported decisions do not automatically remove individuals' legal rights to information, contestation or appropriate review where applicable.
  3. Sensitive data can trigger heightened restrictions and requires particular attention to necessity, safeguards and permitted uses.
  4. Some AI uses face outright restrictions; labeling a system innovative does not resolve an applicable prohibition.

Explanation

Correct Answer

A:

For the technical explanation, collecting more personal data than required increases exposure without automatically improving the model. That is why the case calls for this action: remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient

Incorrect Answers

B:

That explains why a team might design a process to identify relevant data subject requests and assess applicable automated-decision safeguards, but it does not justify the decision required by the case. Collecting more personal data than required increases exposure without automatically improving the model.

C:

For the technical explanation, that explains why a team might escalate biometric or other specially protected personal data for a jurisdiction-specific legal and security review, but it does not justify the decision required by the case. Collecting more personal data than required increases exposure without automatically improving the model

D:

That explains why a team might conduct a jurisdiction- and use-specific prohibited-practice screen before selecting a use case for development, but it does not justify the decision required by the case. Collecting more personal data than required increases exposure without automatically improving the model.

Question 198

The organization has implemented a control for this case: An education provider wants a help-desk classifier and proposes adding pupils' home addresses. Which result would best validate it?

  1. Test sample requests for access, correction and review against the governing jurisdictional rules and operational logs.
  2. Compare approved feature requirements with actual datasets and test a reduced-data version for utility.
  3. Confirm the applicable category, explicit necessity assessment, protective controls and approved processing conditions.
  4. Record the affected people, purpose, jurisdiction, current legal classification and legal disposition.

Explanation

Correct Answer

B:

For the verification, compare approved feature requirements with actual datasets and test a reduced-data version for utility. This verifies the necessary safeguard rather than merely showing that an AI project was completed. Collecting more personal data than required increases exposure without automatically improving the model

Incorrect Answers

A:

Test sample requests for access, correction and review against the governing jurisdictional rules and operational logs would test a different outcome. The directly relevant evidence is to compare approved feature requirements with actual datasets and test a reduced-data version for utility.

C:

For the verification, confirm the applicable category, explicit necessity assessment, protective controls and approved processing conditions would test a different outcome. The directly relevant evidence is to compare approved feature requirements with actual datasets and test a reduced-data version for utility

D:

Record the affected people, purpose, jurisdiction, current legal classification and legal disposition would test a different outcome. The directly relevant evidence is to compare approved feature requirements with actual datasets and test a reduced-data version for utility.

Question 199

The organization has this problem: An education provider wants a help-desk classifier and proposes adding pupils' home addresses. It takes a shortcut: Feed the entire customer database into the model because storage and compute are available. What is the principal risk?

  1. People may lose legally required recourse or receive no usable explanation of a consequential outcome.
  2. An unauthorized new processing purpose can be hidden behind existing access rights.
  3. Unneeded sensitive fields can leak or influence outcomes while adding no justified benefit.
  4. Public outputs can create contractual or infringement exposure even without intentional copying.

Explanation

Correct Answer

C:

For the risk comparison, unneeded sensitive fields can leak or influence outcomes while adding no justified benefit. The shortcut overlooks the central safeguard: remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. Collecting more personal data than required increases exposure without automatically improving the model

Incorrect Answers

A:

This consequence follows more directly from another shortcut: Tell rejected applicants that model decisions can never be reviewed because the system is automated. The shortcut actually described would instead create this risk: Unneeded sensitive fields can leak or influence outcomes while adding no justified benefit.

B:

For the risk comparison, this consequence follows more directly from another shortcut: Assume a dataset may be trained on because staff already can open the underlying system. The shortcut actually described would instead create this risk: Unneeded sensitive fields can leak or influence outcomes while adding no justified benefit

D:

This consequence follows more directly from another shortcut: Assume every generated image is automatically free of third-party rights or usage limitations. The shortcut actually described would instead create this risk: Unneeded sensitive fields can leak or influence outcomes while adding no justified benefit.

Question 200

The governance team needs both a decision and a way to verify it for this case: An education provider wants a help-desk classifier and proposes adding pupils' home addresses. Which pair is best?

  1. Action: Identify and document the applicable lawful basis before reusing personal information for model development. Evidence: Check the data-use assessment for jurisdiction, purpose, applicable lawful basis and approval before ingestion.
  2. Action: Escalate biometric or other specially protected personal data for a jurisdiction-specific legal and security review. Evidence: Confirm the applicable category, explicit necessity assessment, protective controls and approved processing conditions.
  3. Action: Map AI context, measure identified risks and manage prioritized responses using the NIST AI RMF functions. Evidence: Trace a selected use case from contextual mapping to measured evidence and a documented risk response.
  4. Action: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. Evidence: Compare approved feature requirements with actual datasets and test a reduced-data version for utility.

Explanation

Correct Answer

D:

For the action and verification, remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. Compare approved feature requirements with actual datasets and test a reduced-data version for utility. Together, they implement and test the intended safeguard. Collecting more personal data than required increases exposure without automatically improving the model

Incorrect Answers

A:

For the action and verification, this pair supports another governance intervention: Identify and document the applicable lawful basis before reusing personal information for model development. For the actual case, the required action is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. The supporting evidence is: Compare approved feature requirements with actual datasets and test a reduced-data version for utility

B:

This pair supports another governance intervention: Escalate biometric or other specially protected personal data for a jurisdiction-specific legal and security review. For the actual case, the required action is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. The supporting evidence is: Compare approved feature requirements with actual datasets and test a reduced-data version for utility.

C:

This pair supports another governance intervention: Map AI context, measure identified risks and manage prioritized responses using the NIST AI RMF functions. For the actual case, the required action is: Remove unnecessary identifiers and features before training and inference, while testing whether utility remains sufficient. The supporting evidence is: Compare approved feature requirements with actual datasets and test a reduced-data version for utility.

Artificial Intelligence Governance Professional (AIGP) Practice Tests