IAPP AIGP Practice Test 58
Question 570
A school deploys an attendance-risk model with no input from students with disabilities or their advocates. Model assurance reviewers need a documented control before approving continued experimentation. Which specific implementation best addresses the identified gap?
- Insert an effective clinician review and override before consequential orders are committed.
- Conduct an AI impact assessment with accessible stakeholder participation and documented mitigations.
- Prefer eliminating or minimizing the hazard, then add technical barriers, human controls and communication safeguards.
- Write a bounded use-case specification with human authority, exclusions and measurable patient-safety objectives.
Explanation
Correct Answer
B:Conduct an AI impact assessment with accessible stakeholder participation and documented mitigations. Correct. It directly reduces or governs the specific documented risk. Impact assessment should consider people and groups who bear risk, not only the organization buying AI.
Incorrect Answers
A:Insert an effective clinician review and override before consequential orders are committed. Not selected. That action treats another control objective, not the specific gap described. It belongs with Human-in-the-loop design: Human oversight works only when a person can understand, decide and intervene at the right time. The present case instead concerns Stakeholder impact assessment.
C:Prefer eliminating or minimizing the hazard, then add technical barriers, human controls and communication safeguards. Not selected. That action treats another control objective, not the specific gap described. It belongs with Risk mitigation hierarchy: Risk treatment should prioritize effective hazard reduction over disclaimers that shift burden to users. The present case instead concerns Stakeholder impact assessment.
D:Write a bounded use-case specification with human authority, exclusions and measurable patient-safety objectives. Not selected. That action treats another control objective, not the specific gap described. It belongs with Use-case problem framing: A defensible AI design begins with the problem and context, not a preference for a model technology. The present case instead concerns Stakeholder impact assessment.
Question 571
A school deploys an attendance-risk model with no input from students with disabilities or their advocates. Model assurance reviewers need a documented control before approving continued experimentation. Which acceptance test most directly verifies the relevant control?
- Test whether clinicians can reject the suggestion without workflow pressure and log their rationale.
- Use adverse-case tests to show harm reduction from the chosen primary control, not just an acknowledgment banner.
- Measure performance on representative cohorts using the exact proposed production threshold.
- Verify the assessment includes relevant harms, likelihood, severity and accountable treatment actions.
Explanation
Correct Answer
D:Verify the assessment includes relevant harms, likelihood, severity and accountable treatment actions. Correct. It yields measurable evidence that the case-specific safeguard works. Impact assessment should consider people and groups who bear risk, not only the organization buying AI.
Incorrect Answers
A:Test whether clinicians can reject the suggestion without workflow pressure and log their rationale. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Human-in-the-loop design: Human oversight works only when a person can understand, decide and intervene at the right time. The present case instead concerns Stakeholder impact assessment.
B:Use adverse-case tests to show harm reduction from the chosen primary control, not just an acknowledgment banner. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Risk mitigation hierarchy: Risk treatment should prioritize effective hazard reduction over disclaimers that shift burden to users. The present case instead concerns Stakeholder impact assessment.
C:Measure performance on representative cohorts using the exact proposed production threshold. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Thresholds and acceptance metrics: Governance metrics should reflect the actual harm of error types, not aggregate accuracy alone. The present case instead concerns Stakeholder impact assessment.
Question 572
A school deploys an attendance-risk model with no input from students with disabilities or their advocates. Model assurance reviewers need a documented control before approving continued experimentation. For a defensible decision, choose TWO: one diagnosis and one controlled response.
- Assess plausible severity, exposure, uncertainty and probability for foreseeable failure scenarios.
- Identify affected populations, possible disparate impacts, appeal routes and missing stakeholder views.
- Conduct an AI impact assessment with accessible stakeholder participation and documented mitigations.
- Threat-model prompt injection and data exfiltration; segregate instructions, restrict tools and validate outputs.
Explanation
Correct Answers
B:Identify affected populations, possible disparate impacts, appeal routes and missing stakeholder views. Correct. As the first, evidence-gathering half of the requested pair, this establishes the specific risk boundary. Impact assessment should consider people and groups who bear risk, not only the organization buying AI.
C:Conduct an AI impact assessment with accessible stakeholder participation and documented mitigations. Correct. As the second, implementation half of the requested pair, this directly applies the needed safeguard. Impact assessment should consider people and groups who bear risk, not only the organization buying AI.
Incorrect Answers
A:Assess plausible severity, exposure, uncertainty and probability for foreseeable failure scenarios. Not selected. It addresses a different governance mechanism, not the required assessment-and-control pair. It belongs with Risk severity and probability matrix: Low observed frequency in a limited test cannot automatically dismiss a severe AI hazard. The present case instead concerns Stakeholder impact assessment.
D:Threat-model prompt injection and data exfiltration; segregate instructions, restrict tools and validate outputs. Not selected. It addresses a different governance mechanism, not the required assessment-and-control pair. It belongs with Generative AI misuse threat model: Generative AI development requires threat modeling for instruction confusion and downstream actions. The present case instead concerns Stakeholder impact assessment.
Question 573
A school deploys an attendance-risk model with no input from students with disabilities or their advocates. A material training or architecture change makes the previous technical approval insufficient. Which diagnostic task establishes the risk relevant to this described system?
- Identify the decision, affected users, current baseline, proposed intervention and success criteria before selecting a model.
- Trace trust boundaries from uploaded content to model prompts, tool calls and authorized business actions.
- Identify affected populations, possible disparate impacts, appeal routes and missing stakeholder views.
- Find missing architecture decisions, data assumptions, baseline comparisons and review sign-offs.
Explanation
Correct Answer
C:Identify affected populations, possible disparate impacts, appeal routes and missing stakeholder views. Correct. It establishes the relevant factual risk boundary before implementation. Impact assessment should consider people and groups who bear risk, not only the organization buying AI.
Incorrect Answers
A:Identify the decision, affected users, current baseline, proposed intervention and success criteria before selecting a model. Not selected. That check primarily diagnoses a different governance issue. It belongs with Use-case problem framing: A defensible AI design begins with the problem and context, not a preference for a model technology. The present case instead concerns Stakeholder impact assessment.
B:Trace trust boundaries from uploaded content to model prompts, tool calls and authorized business actions. Not selected. That check primarily diagnoses a different governance issue. It belongs with Generative AI misuse threat model: Generative AI development requires threat modeling for instruction confusion and downstream actions. The present case instead concerns Stakeholder impact assessment.
D:Find missing architecture decisions, data assumptions, baseline comparisons and review sign-offs. Not selected. That check primarily diagnoses a different governance issue. It belongs with Design record and decision provenance: Design documentation makes technical tradeoffs governable and defensible across staff changes. The present case instead concerns Stakeholder impact assessment.
Question 574
A school deploys an attendance-risk model with no input from students with disabilities or their advocates. A material training or architecture change makes the previous technical approval insufficient. What is the strongest proportionate corrective design decision?
- Prefer eliminating or minimizing the hazard, then add technical barriers, human controls and communication safeguards.
- Conduct an AI impact assessment with accessible stakeholder participation and documented mitigations.
- Threat-model prompt injection and data exfiltration; segregate instructions, restrict tools and validate outputs.
- Write a bounded use-case specification with human authority, exclusions and measurable patient-safety objectives.
Explanation
Correct Answer
B:Conduct an AI impact assessment with accessible stakeholder participation and documented mitigations. Correct. It directly reduces or governs the specific documented risk. Impact assessment should consider people and groups who bear risk, not only the organization buying AI.
Incorrect Answers
A:Prefer eliminating or minimizing the hazard, then add technical barriers, human controls and communication safeguards. Not selected. That action treats another control objective, not the specific gap described. It belongs with Risk mitigation hierarchy: Risk treatment should prioritize effective hazard reduction over disclaimers that shift burden to users. The present case instead concerns Stakeholder impact assessment.
C:Threat-model prompt injection and data exfiltration; segregate instructions, restrict tools and validate outputs. Not selected. That action treats another control objective, not the specific gap described. It belongs with Generative AI misuse threat model: Generative AI development requires threat modeling for instruction confusion and downstream actions. The present case instead concerns Stakeholder impact assessment.
D:Write a bounded use-case specification with human authority, exclusions and measurable patient-safety objectives. Not selected. That action treats another control objective, not the specific gap described. It belongs with Use-case problem framing: A defensible AI design begins with the problem and context, not a preference for a model technology. The present case instead concerns Stakeholder impact assessment.
Question 575
A school deploys an attendance-risk model with no input from students with disabilities or their advocates. A material training or architecture change makes the previous technical approval insufficient. Which observable evidence best demonstrates the chosen safeguard is working?
- Verify the assessment includes relevant harms, likelihood, severity and accountable treatment actions.
- Pilot representative workflows and compare outcomes with the agreed non-AI baseline.
- Use adverse-case tests to show harm reduction from the chosen primary control, not just an acknowledgment banner.
- Challenge the system with adversarial documents and confirm no secret disclosure or unauthorized tool action.
Explanation
Correct Answer
A:Verify the assessment includes relevant harms, likelihood, severity and accountable treatment actions. Correct. It yields measurable evidence that the case-specific safeguard works. Impact assessment should consider people and groups who bear risk, not only the organization buying AI.
Incorrect Answers
B:Pilot representative workflows and compare outcomes with the agreed non-AI baseline. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Use-case problem framing: A defensible AI design begins with the problem and context, not a preference for a model technology. The present case instead concerns Stakeholder impact assessment.
C:Use adverse-case tests to show harm reduction from the chosen primary control, not just an acknowledgment banner. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Risk mitigation hierarchy: Risk treatment should prioritize effective hazard reduction over disclaimers that shift burden to users. The present case instead concerns Stakeholder impact assessment.
D:Challenge the system with adversarial documents and confirm no secret disclosure or unauthorized tool action. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Generative AI misuse threat model: Generative AI development requires threat modeling for instruction confusion and downstream actions. The present case instead concerns Stakeholder impact assessment.
Question 576
A school deploys an attendance-risk model with no input from students with disabilities or their advocates. A material training or architecture change makes the previous technical approval insufficient. A material exception remains unresolved. Which containment decision best protects affected parties?
- Pause expansion until material affected-group impacts receive assessment and response.
- Do not approve development scope until benefits, user responsibilities and safety constraints are defined.
- Reopen high-impact design decisions lacking material evidence before final approval.
- Identify the decision, affected users, current baseline, proposed intervention and success criteria before selecting a model.
Explanation
Correct Answer
A:Pause expansion until material affected-group impacts receive assessment and response. Correct. It limits further reliance or exposure while the risk remains unresolved. Impact assessment should consider people and groups who bear risk, not only the organization buying AI.
Incorrect Answers
B:Do not approve development scope until benefits, user responsibilities and safety constraints are defined. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Use-case problem framing: A defensible AI design begins with the problem and context, not a preference for a model technology. The present case instead concerns Stakeholder impact assessment.
C:Reopen high-impact design decisions lacking material evidence before final approval. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Design record and decision provenance: Design documentation makes technical tradeoffs governable and defensible across staff changes. The present case instead concerns Stakeholder impact assessment.
D:Identify the decision, affected users, current baseline, proposed intervention and success criteria before selecting a model. Not selected. This action relates to Use-case problem framing. A defensible AI design begins with the problem and context, not a preference for a model technology. For the contain decision, it does not resolve the specifically described issue of Stakeholder impact assessment in A school deploys an attendance-risk model with no input from students with disabilities or their advocates. A material training or architecture change makes the previous technical approval insufficient.
Question 577
A logistics company ranks model risks using only how often failures happened in a small pilot, ignoring severity of a catastrophic collision. The development team must show evidence before promoting the next model version. Which diagnostic task establishes the risk relevant to this described system?
- Find missing architecture decisions, data assumptions, baseline comparisons and review sign-offs.
- Assess plausible severity, exposure, uncertainty and probability for foreseeable failure scenarios.
- Examine the actual intervention point, reviewer workload, authority and automation-bias risk.
- Determine whether the harmful use can be avoided or redesigned before relying on disclosure and warnings.
Explanation
Correct Answer
B:Assess plausible severity, exposure, uncertainty and probability for foreseeable failure scenarios. Correct. It establishes the relevant factual risk boundary before implementation. Low observed frequency in a limited test cannot automatically dismiss a severe AI hazard.
Incorrect Answers
A:Find missing architecture decisions, data assumptions, baseline comparisons and review sign-offs. Not selected. That check primarily diagnoses a different governance issue. It belongs with Design record and decision provenance: Design documentation makes technical tradeoffs governable and defensible across staff changes. The present case instead concerns Risk severity and probability matrix.
C:Examine the actual intervention point, reviewer workload, authority and automation-bias risk. Not selected. That check primarily diagnoses a different governance issue. It belongs with Human-in-the-loop design: Human oversight works only when a person can understand, decide and intervene at the right time. The present case instead concerns Risk severity and probability matrix.
D:Determine whether the harmful use can be avoided or redesigned before relying on disclosure and warnings. Not selected. That check primarily diagnoses a different governance issue. It belongs with Risk mitigation hierarchy: Risk treatment should prioritize effective hazard reduction over disclaimers that shift burden to users. The present case instead concerns Risk severity and probability matrix.
Question 578
A logistics company ranks model risks using only how often failures happened in a small pilot, ignoring severity of a catastrophic collision. The development team must show evidence before promoting the next model version. What is the strongest proportionate corrective design decision?
- Write a bounded use-case specification with human authority, exclusions and measurable patient-safety objectives.
- Maintain a versioned design record of purpose, alternatives, constraints, risk choices and responsible approvers.
- Conduct an AI impact assessment with accessible stakeholder participation and documented mitigations.
- Apply a documented likelihood-severity methodology and higher safeguards for rare but severe harms.
Explanation
Correct Answer
D:Apply a documented likelihood-severity methodology and higher safeguards for rare but severe harms. Correct. It directly reduces or governs the specific documented risk. Low observed frequency in a limited test cannot automatically dismiss a severe AI hazard.
Incorrect Answers
A:Write a bounded use-case specification with human authority, exclusions and measurable patient-safety objectives. Not selected. That action treats another control objective, not the specific gap described. It belongs with Use-case problem framing: A defensible AI design begins with the problem and context, not a preference for a model technology. The present case instead concerns Risk severity and probability matrix.
B:Maintain a versioned design record of purpose, alternatives, constraints, risk choices and responsible approvers. Not selected. That action treats another control objective, not the specific gap described. It belongs with Design record and decision provenance: Design documentation makes technical tradeoffs governable and defensible across staff changes. The present case instead concerns Risk severity and probability matrix.
C:Conduct an AI impact assessment with accessible stakeholder participation and documented mitigations. Not selected. That action treats another control objective, not the specific gap described. It belongs with Stakeholder impact assessment: Impact assessment should consider people and groups who bear risk, not only the organization buying AI. The present case instead concerns Risk severity and probability matrix.
Question 579
A logistics company ranks model risks using only how often failures happened in a small pilot, ignoring severity of a catastrophic collision. The development team must show evidence before promoting the next model version. Which observable evidence best demonstrates the chosen safeguard is working?
- Test whether clinicians can reject the suggestion without workflow pressure and log their rationale.
- Pilot representative workflows and compare outcomes with the agreed non-AI baseline.
- Challenge the system with adversarial documents and confirm no secret disclosure or unauthorized tool action.
- Re-score representative edge cases and confirm treatment priority follows the approved risk rubric.
Explanation
Correct Answer
D:Re-score representative edge cases and confirm treatment priority follows the approved risk rubric. Correct. It yields measurable evidence that the case-specific safeguard works. Low observed frequency in a limited test cannot automatically dismiss a severe AI hazard.
Incorrect Answers
A:Test whether clinicians can reject the suggestion without workflow pressure and log their rationale. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Human-in-the-loop design: Human oversight works only when a person can understand, decide and intervene at the right time. The present case instead concerns Risk severity and probability matrix.
B:Pilot representative workflows and compare outcomes with the agreed non-AI baseline. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Use-case problem framing: A defensible AI design begins with the problem and context, not a preference for a model technology. The present case instead concerns Risk severity and probability matrix.
C:Challenge the system with adversarial documents and confirm no secret disclosure or unauthorized tool action. Not selected. That test would not demonstrate this particular risk has been addressed. It belongs with Generative AI misuse threat model: Generative AI development requires threat modeling for instruction confusion and downstream actions. The present case instead concerns Risk severity and probability matrix.
Question 580
A logistics company ranks model risks using only how often failures happened in a small pilot, ignoring severity of a catastrophic collision. The development team must show evidence before promoting the next model version. A material exception remains unresolved. Which containment decision best protects affected parties?
- Route high-uncertainty cases to review and defer release until accepted error costs are met.
- Stop high-severity deployments when credible harms exceed accepted residual-risk limits.
- Disable the harmful capability if residual risk remains beyond tolerance after feasible mitigation.
- Disable auto-ordering where meaningful clinical oversight is absent.
Explanation
Correct Answer
B:Stop high-severity deployments when credible harms exceed accepted residual-risk limits. Correct. It limits further reliance or exposure while the risk remains unresolved. Low observed frequency in a limited test cannot automatically dismiss a severe AI hazard.
Incorrect Answers
A:Route high-uncertainty cases to review and defer release until accepted error costs are met. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Thresholds and acceptance metrics: Governance metrics should reflect the actual harm of error types, not aggregate accuracy alone. The present case instead concerns Risk severity and probability matrix.
C:Disable the harmful capability if residual risk remains beyond tolerance after feasible mitigation. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Risk mitigation hierarchy: Risk treatment should prioritize effective hazard reduction over disclaimers that shift burden to users. The present case instead concerns Risk severity and probability matrix.
D:Disable auto-ordering where meaningful clinical oversight is absent. Not selected. That response contains a different failure mode and leaves the stated risk unresolved. It belongs with Human-in-the-loop design: Human oversight works only when a person can understand, decide and intervene at the right time. The present case instead concerns Risk severity and probability matrix.