IAPP AIGP Practice Test 57 - Prepaway
Practice Exams:

Pass IAPP AIGP Exam in First Attempt Guaranteed!

Get 100% Latest Exam Questions, Accurate & Verified Answers!

30 Days Free Updates, Instant Download!

Verified By Experts
Premium Bundle
$29.98
$79.99 $109.97

AIGP Premium Bundle


  • Premium File 222 Questions & Answers. Last update: Nov 15, 2026
  • Training Course 124 Lectures
  • Study Guide 566 Pages

IAPP AIGP Practice Test 57

Question 560

A model team has many architecture slides but cannot explain how a risky input feature was approved. Which governance intervention most directly addresses this specific situation?

  1. Consider removing the hazardous capability or limiting access before relying on warnings.
  2. Compare candidate architectures against task complexity, interpretability, cost and evidence requirements.
  3. Record data, model, objective and risk decisions with owners and their supporting evidence.
  4. Add subgroup error thresholds and a remediation plan to the design acceptance criteria.

Explanation

Correct Answer

C:

Record data, model, goal and risk decisions with owners and their supporting evidence. Traceability makes design choices accountable and supports change impact analysis.

Incorrect Answers

A:

Consider removing the hazardous capability or limiting access before relying on warnings. Traceability makes design choices accountable and supports change impact analysis. Record data, model, goal and risk decisions with owners and their supporting evidence.

B:

Compare candidate architectures against task complexity, interpretability, cost and evidence requirements. Traceability makes design choices accountable and supports change impact analysis. Record data, model, goal and risk decisions with owners and their supporting evidence.

D:

Add subgroup error thresholds and a remediation plan to the design acceptance criteria. Traceability makes design choices accountable and supports change impact analysis. Record data, model, goal and risk decisions with owners and their supporting evidence.

Question 561

A retail forecast model abruptly degrades after a data pipeline changes units from dollars to cents. Which governance intervention most directly addresses this specific situation?

  1. Assess memorization and re-identification risk before sharing synthetic data.
  2. Version code, dependencies, datasets and evaluation settings for reproducible experiments.
  3. Add schema, range and distribution validation at ingestion with lineage and rollback.
  4. Measure calibration alongside discrimination and correct or qualify probability estimates.

Explanation

Correct Answer

C:

Add schema, range and distribution validation at ingestion with lineage and rollback. The action addresses training data integrity checks rather than treating every AI control as interchangeable. The decisive signal to verify is that feature values jump by a factor of 100 without a corresponding real business change. This matters because data integrity checks are essential governance controls for model training and retraining. Success should be measured by confirming that invalid batches are blocked and recovery replays the last accepted data version.

Incorrect Answers

A:

Assess memorization and re-identification risk before sharing synthetic data. Data integrity checks are essential governance controls for model training and retraining. Add schema, range and distribution validation at ingestion with lineage and rollback.

B:

Version code, dependencies, datasets and evaluation settings for reproducible experiments. Data integrity checks are essential governance controls for model training and retraining. Add schema, range and distribution validation at ingestion with lineage and rollback.

D:

Measure calibration alongside discrimination and correct or qualify probability estimates. Data integrity checks are essential governance controls for model training and retraining. Add schema, range and distribution validation at ingestion with lineage and rollback.

Question 562

A design agency plans to sell generated illustrations without reviewing whether source asset licenses permit the intended commercialization. Which governance intervention most directly addresses this specific situation?

  1. Review lending nondiscrimination and decision-notice obligations appropriate to the jurisdiction.
  2. Review source-license restrictions and output rights before commercial distribution.
  3. Review insurance fairness rules and model factors with legal and actuarial oversight.
  4. Evaluate applicable product-safety and liability requirements alongside rigorous fault testing.

Explanation

Correct Answer

B:

Review source-license restrictions and output rights before commercial distribution. The action addresses output licensing and derivative works rather than treating every AI control as interchangeable. The decisive signal to verify is that the generation workflow relies on assets labeled noncommercial-only in a supplied content library. This matters because generative output does not automatically erase rights and restrictions attached to underlying assets. Success should be measured by confirming that the approved content workflow documents legal use rights for all material inputs and planned distribution.

Incorrect Answers

A:

Review lending nondiscrimination and decision-notice obligations appropriate to the jurisdiction. Generative output does not automatically erase rights and restrictions attached to underlying assets. Review source-license restrictions and output rights before commercial distribution.

C:

Review insurance fairness rules and model factors with legal and actuarial oversight. Generative output does not automatically erase rights and restrictions attached to underlying assets. Review source-license restrictions and output rights before commercial distribution.

D:

Evaluate applicable product-safety and liability requirements alongside rigorous fault testing. Generative output does not automatically erase rights and restrictions attached to underlying assets. Review source-license restrictions and output rights before commercial distribution.

Question 563

A science lab cannot reproduce a fairness metric because training code, random seeds and dataset versions were not retained. Which governance reasoning best supports the appropriate response?

  1. Sensitive information may be inferred by AI even when the original dataset appears ordinary.
  2. Synthetic generation does not by itself guarantee privacy or lawful disclosure.
  3. Data integrity checks are essential governance controls for model training and retraining.
  4. Reproducibility turns an evaluation assertion into testable evidence.

Explanation

Correct Answer

D:

Reproducibility turns an evaluation assertion into testable evidence. This reasoning fits reproducibility of model experiments because the observed gap is that two analysts obtain incompatible results despite citing the same report title. It points to the practical intervention: Version code, dependencies, datasets and evaluation settings for reproducible experiments. The outcome should then show that a third party can reproduce the released model's evaluation within declared tolerances.

Incorrect Answers

A:

Sensitive information may be inferred by AI even when the original dataset appears ordinary. This reasoning fits reproducibility of model experiments because the observed gap is that two analysts obtain incompatible results despite citing the same report title. Reproducibility turns an evaluation assertion into testable evidence.

B:

Synthetic generation does not by itself guarantee privacy or lawful disclosure. This reasoning fits reproducibility of model experiments because the observed gap is that two analysts obtain incompatible results despite citing the same report title. Reproducibility turns an evaluation assertion into testable evidence.

C:

Data integrity checks are essential governance controls for model training and retraining. This reasoning fits reproducibility of model experiments because the observed gap is that two analysts obtain incompatible results despite citing the same report title. Reproducibility turns an evaluation assertion into testable evidence.

Question 564

A service provider must explain its AI system's intended use and known limits to client deployers before launch. Select the ONE intervention-and-verification pair that directly resolves this specific governance gap.

  1. Intervention: Create accurate technical and operational disclosures linked to the released model version. / Verification: Downstream users receive current limitation and oversight guidance with the actual release.
  2. Intervention: Create accurate technical and operational disclosures linked to the released model version. / Verification: The approved release record links version, tests, known limitations and responsible owners.
  3. Intervention: Require root-cause investigation, test evidence, owner and verified closure for model issues. / Verification: Downstream users receive current limitation and oversight guidance with the actual release.
  4. Intervention: Set harm-specific monitoring thresholds and an alert-to-containment workflow. / Verification: Threshold breaches produce timely alerts and tested containment decisions.

Explanation

Correct Answer

A:

Intervention: Create accurate technical and operational disclosures linked to the released model version. / Verification: Downstream users receive current limitation and oversight guidance with the actual release. The reason to require this combination is that transparency requires actionable deployment information, not just generic AI branding.

Incorrect Answers

B:

Intervention: Create accurate technical and operational disclosures linked to the released model version. / Verification: The approved release record links version, tests, known limitations and responsible owners. The reason to require this combination is that transparency requires actionable deployment information, not just generic AI branding. Intervention: Create accurate technical and operational disclosures linked to the released model version. / Verification: Downstream users receive current limitation and oversight guidance with the actual release.

C:

Intervention: Require root-cause investigation, test evidence, owner and verified closure for model issues. / Verification: Downstream users receive current limitation and oversight guidance with the actual release. The reason to require this combination is that transparency requires actionable deployment information, not just generic AI branding. Intervention: Create accurate technical and operational disclosures linked to the released model version. / Verification: Downstream users receive current limitation and oversight guidance with the actual release.

D:

Intervention: Set harm-specific monitoring thresholds and an alert-to-containment workflow. / Verification: Threshold breaches produce timely alerts and tested containment decisions. The reason to require this combination is that transparency requires actionable deployment information, not just generic AI branding. Intervention: Create accurate technical and operational disclosures linked to the released model version. / Verification: Downstream users receive current limitation and oversight guidance with the actual release.

Question 565

A publisher finds an AI startup scraped subscription articles for training despite license terms restricting automated reuse. Which governance reasoning best supports the appropriate response?

  1. Generative output does not automatically erase rights and restrictions attached to underlying assets.
  2. Indirect features may reproduce prohibited discrimination even without explicit protected traits.
  3. Public accessibility does not necessarily grant a license for every AI training use.
  4. AI-assisted employment decisions remain subject to discrimination rules even when scores are automated.

Explanation

Correct Answer

C:

Public accessibility does not necessarily grant a license for every AI training use. This reasoning fits licensed training content because the observed gap is that dataset manifests identify paid articles whose licenses exclude the proposed training purpose. It points to the practical intervention: Assess copyright and license permissions and remove or remediate unauthorized training use. The outcome should then show that the training-data register contains documented rights or excludes nonpermitted works.

Incorrect Answers

A:

Generative output does not automatically erase rights and restrictions attached to underlying assets. This reasoning fits licensed training content because the observed gap is that dataset manifests identify paid articles whose licenses exclude the proposed training purpose. Public accessibility does not necessarily grant a license for every AI training use.

B:

Indirect features may reproduce prohibited discrimination even without explicit protected traits. This reasoning fits licensed training content because the observed gap is that dataset manifests identify paid articles whose licenses exclude the proposed training purpose. Public accessibility does not necessarily grant a license for every AI training use.

D:

AI-assisted employment decisions remain subject to discrimination rules even when scores are automated. This reasoning fits licensed training content because the observed gap is that dataset manifests identify paid articles whose licenses exclude the proposed training purpose. Public accessibility does not necessarily grant a license for every AI training use.

Question 566

A model-hosting team monitors API uptime but never reviews model artifact tampering or dependency compromise. Select the ONE intervention-and-verification pair that directly resolves this specific governance gap.

  1. Intervention: Threat-model the artifact, build pipeline and inference dependencies and add integrity controls. / Verification: Downstream users receive current limitation and oversight guidance with the actual release.
  2. Intervention: Investigate drift with data, product and operational experts before changing infrastructure. / Verification: Only approved, integrity-verified artifacts are promoted and alerts cover tampering attempts.
  3. Intervention: Reassess risk, owners and residual exposure whenever material features or sources change. / Verification: Updated assessments and mitigation owners accompany the new version before release.
  4. Intervention: Threat-model the artifact, build pipeline and inference dependencies and add integrity controls. / Verification: Only approved, integrity-verified artifacts are promoted and alerts cover tampering attempts.

Explanation

Correct Answer

D:

Intervention: Threat-model the artifact, build pipeline and inference dependencies and add integrity controls. / Verification: Only approved, integrity-verified artifacts are promoted and alerts cover tampering attempts.

Incorrect Answers

A:

Intervention: Threat-model the artifact, build pipeline and inference dependencies and add integrity controls. / Verification: Downstream users receive current limitation and oversight guidance with the actual release. Intervention: Threat-model the artifact, build pipeline and inference dependencies and add integrity controls. / Verification: Only approved, integrity-verified artifacts are promoted and alerts cover tampering attempts.

B:

Intervention: Investigate drift with data, product and operational experts before changing infrastructure. / Verification: Only approved, integrity-verified artifacts are promoted and alerts cover tampering attempts. Intervention: Threat-model the artifact, build pipeline and inference dependencies and add integrity controls. / Verification: Only approved, integrity-verified artifacts are promoted and alerts cover tampering attempts.

C:

Intervention: Reassess risk, owners and residual exposure whenever material features or sources change. / Verification: Updated assessments and mitigation owners accompany the new version before release. Intervention: Threat-model the artifact, build pipeline and inference dependencies and add integrity controls. / Verification: Only approved, integrity-verified artifacts are promoted and alerts cover tampering attempts.

Question 567

A demand predictor is retrained every Friday without checking whether data changed or validation improved. Select the ONE intervention-and-verification pair that directly resolves this specific governance gap.

  1. Intervention: Set harm-specific monitoring thresholds and an alert-to-containment workflow. / Verification: Threshold breaches produce timely alerts and tested containment decisions.
  2. Intervention: Require root-cause investigation, test evidence, owner and verified closure for model issues. / Verification: Only validated retrained models pass release gates; unneeded changes are skipped.
  3. Intervention: Use an evidence-driven maintenance schedule with material-change triggers and acceptance checks. / Verification: The approved release record links version, tests, known limitations and responsible owners.
  4. Intervention: Use an evidence-driven maintenance schedule with material-change triggers and acceptance checks. / Verification: Only validated retrained models pass release gates; unneeded changes are skipped.

Explanation

Correct Answer

D:

Intervention: Use an evidence-driven maintenance schedule with material-change triggers and acceptance checks. / Verification: Only validated retrained models pass release gates; unneeded changes are skipped. The reason to require this combination is that retraining cadence should respond to model condition and risk rather than calendar habit alone.

Incorrect Answers

A:

Intervention: Set harm-specific monitoring thresholds and an alert-to-containment workflow. / Verification: Threshold breaches produce timely alerts and tested containment decisions. The reason to require this combination is that retraining cadence should respond to model condition and risk rather than calendar habit alone. Intervention: Use an evidence-driven maintenance schedule with material-change triggers and acceptance checks. / Verification: Only validated retrained models pass release gates; unneeded changes are skipped.

B:

Intervention: Require root-cause investigation, test evidence, owner and verified closure for model issues. / Verification: Only validated retrained models pass release gates; unneeded changes are skipped. The reason to require this combination is that retraining cadence should respond to model condition and risk rather than calendar habit alone. Intervention: Use an evidence-driven maintenance schedule with material-change triggers and acceptance checks. / Verification: Only validated retrained models pass release gates; unneeded changes are skipped.

C:

Intervention: Use an evidence-driven maintenance schedule with material-change triggers and acceptance checks. / Verification: The approved release record links version, tests, known limitations and responsible owners. The reason to require this combination is that retraining cadence should respond to model condition and risk rather than calendar habit alone. Intervention: Use an evidence-driven maintenance schedule with material-change triggers and acceptance checks. / Verification: Only validated retrained models pass release gates; unneeded changes are skipped.

Question 568

A utility uses an AI leak detector but only checks whether its inference server stays online. Which test result would best demonstrate that the actual governance concern has been addressed?

  1. A controlled test shows substantive reviewer decisions can change the outcome before final action.
  2. Authorized staff can continue critical processing without using unverified model outputs.
  3. Affected users receive accurate notice, support channels and correction updates.
  4. Monitoring detects declining task performance and routes action before risk tolerance is exceeded.

Explanation

Correct Answer

D:

Monitoring detects declining task performance and routes action before risk tolerance is exceeded. The required intervention is to monitor error patterns, drift and harmful outcomes against operational thresholds. The particular diagnostic issue is that the service reports 99.9 percent availability while field teams miss increasing leak events. The governing distinction is that operational success depends on the impact of predictions, not merely uptime.

Incorrect Answers

A:

A controlled test shows substantive reviewer decisions can change the outcome before final action. Operational success depends on the impact of predictions, not merely uptime. Monitoring detects declining task performance and routes action before risk tolerance is exceeded.

B:

Authorized staff can continue critical processing without using unverified model outputs. Operational success depends on the impact of predictions, not merely uptime. Monitoring detects declining task performance and routes action before risk tolerance is exceeded.

C:

Affected users receive accurate notice, support channels and correction updates. Operational success depends on the impact of predictions, not merely uptime. Monitoring detects declining task performance and routes action before risk tolerance is exceeded.

Question 569

A health insurer's AI segments customers into price groups but management cannot demonstrate the absence of unlawful rating factors. Which additional observation would most strongly confirm the particular risk or control gap?

  1. The contract permits broad provider retention and use beyond the engineering service.
  2. Dataset manifests identify paid articles whose licenses exclude the proposed training purpose.
  3. Pricing models rely on opaque correlations that cannot be linked to permitted risk variables.
  4. Rejected borrowers receive a generic algorithmic score with no verified decision rationale.

Explanation

Correct Answer

C:

Pricing models rely on opaque correlations that cannot be linked to permitted risk variables. This is the diagnostic observation that distinguishes insurance pricing transparency from other concerns within understand how other types of existing laws apply to ai. It supports the targeted response: Review insurance fairness rules and model factors with legal and actuarial oversight. The underlying governance reason is that an insurance model's technical accuracy does not guarantee its rates are legally permissible. A meaningful follow-up would be to verify that the insurer documents permitted rating factors and tests for prohibited proxy effects.

Incorrect Answers

A:

The contract permits broad provider retention and use beyond the engineering service. Pricing models rely on opaque correlations that cannot be linked to permitted risk variables.

B:

Dataset manifests identify paid articles whose licenses exclude the proposed training purpose. Pricing models rely on opaque correlations that cannot be linked to permitted risk variables.

D:

Rejected borrowers receive a generic algorithmic score with no verified decision rationale. Pricing models rely on opaque correlations that cannot be linked to permitted risk variables.

Question 570

An employment service corrects a client's record but its risk-scoring system still uses the stale feature snapshot. Which test result would best demonstrate that the actual governance concern has been addressed?

  1. The input dataset and model pipeline exclude unnecessary medical fields and document the rationale.
  2. A correction test updates downstream data and triggers appropriate review of affected decisions.
  3. A documented lawful-purpose assessment determines permitted training use and any required additional notices or safeguards.
  4. Signed terms allocate data-protection roles and operational tests confirm instruction-bound processing.

Explanation

Correct Answer

B:

A correction test updates downstream data and triggers appropriate review of affected decisions. Data-subject rights require attention to AI-related copies and downstream processing, not just front-end records.

Incorrect Answers

A:

The input dataset and model pipeline exclude unnecessary medical fields and document the rationale. Data-subject rights require attention to AI-related copies and downstream processing, not just front-end records. A correction test updates downstream data and triggers appropriate review of affected decisions.

C:

A documented lawful-purpose assessment determines permitted training use and any required additional notices or safeguards. Data-subject rights require attention to AI-related copies and downstream processing, not just front-end records. A correction test updates downstream data and triggers appropriate review of affected decisions.

D:

Signed terms allocate data-protection roles and operational tests confirm instruction-bound processing. Data-subject rights require attention to AI-related copies and downstream processing, not just front-end records. A correction test updates downstream data and triggers appropriate review of affected decisions.

Artificial Intelligence Governance Professional (AIGP) Practice Tests