ISACA Advanced in AI Audit AAIA Practice Test 4
Question 30
In a review of AI Models, Considerations, and Requirements, investigators cannot establish the expected control proof: p95/p99 traces, failover drills and completed acceptance results against approved slas. The underlying event was that a real-time scoring API occasionally exceeds a response deadline but has excellent average accuracy. Which control weakness is the MOST likely cause requiring focused audit work?
- Nonfunctional latency and availability requirements were excluded from model suitability
- Business loss asymmetry is missing from model acceptance criteria
- Generative probabilistic output is being mistaken for verified authoritative advice
- Normal operational context and false-positive tolerance are absent from design assumptions
Explanation
Correct Answer
A:Correct. The observed condition specifically indicates nonfunctional latency and availability requirements were excluded from model suitability. A meaningful audit response is to verify this diagnosis and then recommend define tail-latency, fail-safe response and reliability service objectives in selection gates.
Incorrect Answers
B:“Business loss asymmetry is missing from model acceptance criteria” concerns a separate audit weakness, not the missing assurance described by nonfunctional latency and availability requirements were excluded from model suitability. The correct cause is “Nonfunctional latency and availability requirements were excluded from model suitability”.
C:“Generative probabilistic output is being mistaken for verified authoritative advice” concerns a separate audit weakness, not the missing assurance described by nonfunctional latency and availability requirements were excluded from model suitability. The correct cause is “Nonfunctional latency and availability requirements were excluded from model suitability”.
D:“Normal operational context and false-positive tolerance are absent from design assumptions” concerns a separate audit weakness, not the missing assurance described by nonfunctional latency and availability requirements were excluded from model suitability. The correct cause is “Nonfunctional latency and availability requirements were excluded from model suitability”.
Question 31
An autonomous purchasing agent can initiate payments but requirements describe it as an informational chatbot. From an AI audit perspective, which corrective action is the MOST direct response to the weakness described, rather than a useful control for some different AI risk?
- Require rolling-origin validation with strict time cutoffs and a final unseen period
- Define image quality, preprocessing, provenance and clinical operating boundaries in system requirements
- Test speech models on actual acoustic environments, accents and supported devices
- Specify tool permissions, spend limits, approvals and rollback semantics for every autonomous action
Explanation
Correct Answer
D:Correct. Specify tool permissions, spend limits, approvals and rollback semantics for every autonomous action is aimed at the actual failure: delegated action authority and transaction boundaries were omitted. The audit should subsequently obtain approved tool-use matrix and negative tests for unauthorized purchase requests to test execution, not merely the existence of a policy.
Incorrect Answers
A:“Require rolling-origin validation with strict time cutoffs and a final unseen period” is a control for temporal leakage from future periods contaminates evaluation, but the audit finding here is delegated action authority and transaction boundaries were omitted. The relevant measure is “Specify tool permissions, spend limits, approvals and rollback semantics for every autonomous action”.
B:“Define image quality, preprocessing, provenance and clinical operating boundaries in system requirements” is a control for input modality, preprocessing and clinical context requirements were omitted, but the audit finding here is delegated action authority and transaction boundaries were omitted. The relevant measure is “Specify tool permissions, spend limits, approvals and rollback semantics for every autonomous action”.
C:“Test speech models on actual acoustic environments, accents and supported devices” is a control for deployment conditions differ from the representative validation environment, but the audit finding here is delegated action authority and transaction boundaries were omitted. The relevant measure is “Specify tool permissions, spend limits, approvals and rollback semantics for every autonomous action”.
Question 32
In an audit of AI Models, Considerations, and Requirements, managers say this measure is live: Specify tool permissions, spend limits, approvals and rollback semantics for every autonomous action. They claim it corrected a material audit gap. What direct operating evidence would allow the auditor to substantiate that particular claim across a defensible sample?
- Dated split manifests and out-of-time backtests showing no future records in features
- Signed multimodal interface specification and testing for unreadable or out-of-scope images
- Approved tool-use matrix and negative tests for unauthorized purchase requests
- Field-stratified word error reports tied to approved operational service thresholds
Explanation
Correct Answer
C:Correct. The proposed records—approved tool-use matrix and negative tests for unauthorized purchase requests—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is delegated action authority and transaction boundaries were omitted.
Incorrect Answers
A:“Dated split manifests and out-of-time backtests showing no future records in features” would test temporal leakage from future periods contaminates evaluation, not the present safeguard against delegated action authority and transaction boundaries were omitted. The expected audit evidence is “Approved tool-use matrix and negative tests for unauthorized purchase requests”.
B:“Signed multimodal interface specification and testing for unreadable or out-of-scope images” would test input modality, preprocessing and clinical context requirements were omitted, not the present safeguard against delegated action authority and transaction boundaries were omitted. The expected audit evidence is “Approved tool-use matrix and negative tests for unauthorized purchase requests”.
D:“Field-stratified word error reports tied to approved operational service thresholds” would test deployment conditions differ from the representative validation environment, not the present safeguard against delegated action authority and transaction boundaries were omitted. The expected audit evidence is “Approved tool-use matrix and negative tests for unauthorized purchase requests”.
Question 33
In a review of AI Models, Considerations, and Requirements, investigators cannot establish the expected control proof: approved tool-use matrix and negative tests for unauthorized purchase requests. The underlying event was that an autonomous purchasing agent can initiate payments but requirements describe it as an informational chatbot. Which control weakness is the MOST likely cause requiring focused audit work?
- Temporal leakage from future periods contaminates evaluation
- Delegated action authority and transaction boundaries were omitted
- Input modality, preprocessing and clinical context requirements were omitted
- Deployment conditions differ from the representative validation environment
Explanation
Correct Answer
B:Correct. The observed condition specifically indicates delegated action authority and transaction boundaries were omitted. A meaningful audit response is to verify this diagnosis and then recommend specify tool permissions, spend limits, approvals and rollback semantics for every autonomous action.
Incorrect Answers
A:“Temporal leakage from future periods contaminates evaluation” concerns a separate audit weakness, not the missing assurance described by delegated action authority and transaction boundaries were omitted. The correct cause is “Delegated action authority and transaction boundaries were omitted”.
C:“Input modality, preprocessing and clinical context requirements were omitted” concerns a separate audit weakness, not the missing assurance described by delegated action authority and transaction boundaries were omitted. The correct cause is “Delegated action authority and transaction boundaries were omitted”.
D:“Deployment conditions differ from the representative validation environment” concerns a separate audit weakness, not the missing assurance described by delegated action authority and transaction boundaries were omitted. The correct cause is “Delegated action authority and transaction boundaries were omitted”.
Question 34
Multiple divisions buy AI subscriptions through corporate cards and no risk officer can list active integrations. From an AI audit perspective, which corrective action is the MOST direct response to the weakness described, rather than a useful control for some different AI risk?
- Maintain a mandatory AI-use register with discovery, ownership and approval workflows
- Require explicit risk-owner approval after independent validation and control exception review
- Adopt balanced benefits, safety, fairness, quality and human-override scorecards
- Establish mapped controls, accountable owners, due dates and documented exception expiry
Explanation
Correct Answer
A:Correct. Maintain a mandatory AI-use register with discovery, ownership and approval workflows is aimed at the actual failure: ai inventory and centralized intake ownership are undefined. The audit should subsequently obtain reconciled expense, network and approved ai inventory with exceptions closed to test execution, not merely the existence of a policy.
Incorrect Answers
B:“Require explicit risk-owner approval after independent validation and control exception review” is a control for technical gate was confused with business acceptance of residual AI risk, but the audit finding here is aI inventory and centralized intake ownership are undefined. The relevant measure is “Maintain a mandatory AI-use register with discovery, ownership and approval workflows”.
C:“Adopt balanced benefits, safety, fairness, quality and human-override scorecards” is a control for aI program metrics are narrowly financial and fail to monitor trust and risk, but the audit finding here is aI inventory and centralized intake ownership are undefined. The relevant measure is “Maintain a mandatory AI-use register with discovery, ownership and approval workflows”.
D:“Establish mapped controls, accountable owners, due dates and documented exception expiry” is a control for policy statements were not translated into auditable operating standards, but the audit finding here is aI inventory and centralized intake ownership are undefined. The relevant measure is “Maintain a mandatory AI-use register with discovery, ownership and approval workflows”.
Question 35
In an audit of AI Governance and Program Management, managers say this measure is live: Maintain a mandatory AI-use register with discovery, ownership and approval workflows. They claim it corrected a material audit gap. What direct operating evidence would allow the auditor to substantiate that particular claim across a defensible sample?
- Dated risk-acceptance record referencing validation issues and required compensating controls
- Board packs showing trend thresholds, breach actions and management responses
- Control register, owner attestations and time-bounded exception follow-up
- Reconciled expense, network and approved AI inventory with exceptions closed
Explanation
Correct Answer
D:Correct. The proposed records—reconciled expense, network and approved ai inventory with exceptions closed—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is ai inventory and centralized intake ownership are undefined.
Incorrect Answers
A:“Dated risk-acceptance record referencing validation issues and required compensating controls” would test technical gate was confused with business acceptance of residual AI risk, not the present safeguard against aI inventory and centralized intake ownership are undefined. The expected audit evidence is “Reconciled expense, network and approved AI inventory with exceptions closed”.
B:“Board packs showing trend thresholds, breach actions and management responses” would test aI program metrics are narrowly financial and fail to monitor trust and risk, not the present safeguard against aI inventory and centralized intake ownership are undefined. The expected audit evidence is “Reconciled expense, network and approved AI inventory with exceptions closed”.
C:“Control register, owner attestations and time-bounded exception follow-up” would test policy statements were not translated into auditable operating standards, not the present safeguard against aI inventory and centralized intake ownership are undefined. The expected audit evidence is “Reconciled expense, network and approved AI inventory with exceptions closed”.
Question 36
In a review of AI Governance and Program Management, investigators cannot establish the expected control proof: reconciled expense, network and approved ai inventory with exceptions closed. The underlying event was that multiple divisions buy AI subscriptions through corporate cards and no risk officer can list active integrations. Which control weakness is the MOST likely cause requiring focused audit work?
- Technical gate was confused with business acceptance of residual AI risk
- AI program metrics are narrowly financial and fail to monitor trust and risk
- AI inventory and centralized intake ownership are undefined
- Policy statements were not translated into auditable operating standards
Explanation
Correct Answer
C:Correct. The observed condition specifically indicates ai inventory and centralized intake ownership are undefined. A meaningful audit response is to verify this diagnosis and then recommend maintain a mandatory ai-use register with discovery, ownership and approval workflows.
Incorrect Answers
A:“Technical gate was confused with business acceptance of residual AI risk” concerns a separate audit weakness, not the missing assurance described by aI inventory and centralized intake ownership are undefined. The correct cause is “AI inventory and centralized intake ownership are undefined”.
B:“AI program metrics are narrowly financial and fail to monitor trust and risk” concerns a separate audit weakness, not the missing assurance described by aI inventory and centralized intake ownership are undefined. The correct cause is “AI inventory and centralized intake ownership are undefined”.
D:“Policy statements were not translated into auditable operating standards” concerns a separate audit weakness, not the missing assurance described by aI inventory and centralized intake ownership are undefined. The correct cause is “AI inventory and centralized intake ownership are undefined”.
Question 37
An AI steering committee approves models but nobody owns postdeployment adverse consumer outcomes. From an AI audit perspective, which corrective action is the MOST direct response to the weakness described, rather than a useful control for some different AI risk?
- Publish role-specific acceptable-use rules and test staff understanding with real data-handling scenarios
- Assign lifecycle accountability, escalation rights and measurable owner KRIs to each material model
- Define supplier assurance owner, periodic review schedule and evidence-based service obligations
- Create risk-tiered sanctioned pathways with controls and technical discovery of shadow use
Explanation
Correct Answer
B:Correct. Assign lifecycle accountability, escalation rights and measurable owner KRIs to each material model is aimed at the actual failure: accountability for ongoing model harm is not allocated to an empowered business owner. The audit should subsequently obtain governance charter, signed raci and escalations acted upon within service targets to test execution, not merely the existence of a policy.
Incorrect Answers
A:“Publish role-specific acceptable-use rules and test staff understanding with real data-handling scenarios” is a control for awareness content is not aligned with enforceable AI data-use policy, but the audit finding here is accountability for ongoing model harm is not allocated to an empowered business owner. The relevant measure is “Assign lifecycle accountability, escalation rights and measurable owner KRIs to each material model”.
C:“Define supplier assurance owner, periodic review schedule and evidence-based service obligations” is a control for third-party AI governance responsibilities were never formally assigned, but the audit finding here is accountability for ongoing model harm is not allocated to an empowered business owner. The relevant measure is “Assign lifecycle accountability, escalation rights and measurable owner KRIs to each material model”.
D:“Create risk-tiered sanctioned pathways with controls and technical discovery of shadow use” is a control for governance is impractical and fails to offer approved low-risk alternatives, but the audit finding here is accountability for ongoing model harm is not allocated to an empowered business owner. The relevant measure is “Assign lifecycle accountability, escalation rights and measurable owner KRIs to each material model”.
Question 38
In an audit of AI Governance and Program Management, managers say this measure is live: Assign lifecycle accountability, escalation rights and measurable owner KRIs to each material model. They claim it corrected a material audit gap. What direct operating evidence would allow the auditor to substantiate that particular claim across a defensible sample?
- Governance charter, signed RACI and escalations acted upon within service targets
- Policy-to-training trace and assessment results for privileged and frontline users
- Contract control schedule, owner RACI and performed vendor monitoring records
- Approved-use catalog, monitoring trends and remediated unauthorized integrations
Explanation
Correct Answer
A:Correct. The proposed records—governance charter, signed raci and escalations acted upon within service targets—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is accountability for ongoing model harm is not allocated to an empowered business owner.
Incorrect Answers
B:“Policy-to-training trace and assessment results for privileged and frontline users” would test awareness content is not aligned with enforceable AI data-use policy, not the present safeguard against accountability for ongoing model harm is not allocated to an empowered business owner. The expected audit evidence is “Governance charter, signed RACI and escalations acted upon within service targets”.
C:“Contract control schedule, owner RACI and performed vendor monitoring records” would test third-party AI governance responsibilities were never formally assigned, not the present safeguard against accountability for ongoing model harm is not allocated to an empowered business owner. The expected audit evidence is “Governance charter, signed RACI and escalations acted upon within service targets”.
D:“Approved-use catalog, monitoring trends and remediated unauthorized integrations” would test governance is impractical and fails to offer approved low-risk alternatives, not the present safeguard against accountability for ongoing model harm is not allocated to an empowered business owner. The expected audit evidence is “Governance charter, signed RACI and escalations acted upon within service targets”.
Question 39
In a review of AI Governance and Program Management, investigators cannot establish the expected control proof: governance charter, signed raci and escalations acted upon within service targets. The underlying event was that an AI steering committee approves models but nobody owns postdeployment adverse consumer outcomes. Which control weakness is the MOST likely cause requiring focused audit work?
- Awareness content is not aligned with enforceable AI data-use policy
- Third-party AI governance responsibilities were never formally assigned
- Governance is impractical and fails to offer approved low-risk alternatives
- Accountability for ongoing model harm is not allocated to an empowered business owner
Explanation
Correct Answer
D:Correct. The observed condition specifically indicates accountability for ongoing model harm is not allocated to an empowered business owner. A meaningful audit response is to verify this diagnosis and then recommend assign lifecycle accountability, escalation rights and measurable owner kris to each material model.
Incorrect Answers
A:“Awareness content is not aligned with enforceable AI data-use policy” concerns a separate audit weakness, not the missing assurance described by accountability for ongoing model harm is not allocated to an empowered business owner. The correct cause is “Accountability for ongoing model harm is not allocated to an empowered business owner”.
B:“Third-party AI governance responsibilities were never formally assigned” concerns a separate audit weakness, not the missing assurance described by accountability for ongoing model harm is not allocated to an empowered business owner. The correct cause is “Accountability for ongoing model harm is not allocated to an empowered business owner”.
C:“Governance is impractical and fails to offer approved low-risk alternatives” concerns a separate audit weakness, not the missing assurance described by accountability for ongoing model harm is not allocated to an empowered business owner. The correct cause is “Accountability for ongoing model harm is not allocated to an empowered business owner”.
Question 40
A high-impact lending model is launched after an architecture review without documented business risk acceptance. From an AI audit perspective, which corrective action is the MOST direct response to the weakness described, rather than a useful control for some different AI risk?
- Introduce risk-tiered release gates and exception escalation with event-triggered oversight
- Route all material AI services through standardized design, privacy, security and risk intake
- Require explicit risk-owner approval after independent validation and control exception review
- Maintain a mandatory AI-use register with discovery, ownership and approval workflows
Explanation
Correct Answer
C:Correct. Require explicit risk-owner approval after independent validation and control exception review is aimed at the actual failure: technical gate was confused with business acceptance of residual ai risk. The audit should subsequently obtain dated risk-acceptance record referencing validation issues and required compensating controls to test execution, not merely the existence of a policy.
Incorrect Answers
A:“Introduce risk-tiered release gates and exception escalation with event-triggered oversight” is a control for governance cadence and approval thresholds are inconsistent with change velocity, but the audit finding here is technical gate was confused with business acceptance of residual AI risk. The relevant measure is “Require explicit risk-owner approval after independent validation and control exception review”.
B:“Route all material AI services through standardized design, privacy, security and risk intake” is a control for procurement path lacks AI-specific architecture and data integration checkpoints, but the audit finding here is technical gate was confused with business acceptance of residual AI risk. The relevant measure is “Require explicit risk-owner approval after independent validation and control exception review”.
D:“Maintain a mandatory AI-use register with discovery, ownership and approval workflows” is a control for aI inventory and centralized intake ownership are undefined, but the audit finding here is technical gate was confused with business acceptance of residual AI risk. The relevant measure is “Require explicit risk-owner approval after independent validation and control exception review”.