ISACA Advanced in AI Audit AAIA Practice Test 40
Question 390
In a review of Incident Response Management Specific to AI, investigators cannot establish the expected control proof: fail-safe drill results and documented manual approval records. The underlying event was that an AI outage causes applications to bypass human approval checks to maintain throughput. Which control weakness is the MOST likely cause requiring focused audit work?
- Business-continuity mode disables critical decision safeguards
- Root-cause analysis and recurrence prevention are absent from AI incident management
- Model incidents do not trace data lineage and affected derivatives
- AI-related inaccurate output is not captured as a reportable quality incident
Explanation
Correct Answer
A:Correct. The observed condition specifically indicates business-continuity mode disables critical decision safeguards. A meaningful audit response is to verify this diagnosis and then recommend restore manual review controls or pause consequential decisions during degraded mode.
Incorrect Answers
B:“Root-cause analysis and recurrence prevention are absent from AI incident management” concerns a separate audit weakness, not the missing assurance described by business-continuity mode disables critical decision safeguards. The correct cause is “Business-continuity mode disables critical decision safeguards”.
C:“Model incidents do not trace data lineage and affected derivatives” concerns a separate audit weakness, not the missing assurance described by business-continuity mode disables critical decision safeguards. The correct cause is “Business-continuity mode disables critical decision safeguards”.
D:“AI-related inaccurate output is not captured as a reportable quality incident” concerns a separate audit weakness, not the missing assurance described by business-continuity mode disables critical decision safeguards. The correct cause is “Business-continuity mode disables critical decision safeguards”.
Question 391
An adversarial input creates unsafe physical robot commands and incident records capture only API errors. 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?
- Trigger circuit breaker, approved fallback and incident classification based on customer harm
- Revoke agent tools, halt transactions and invoke financial incident coordination
- Link model-provider identifiers to affected services and activate targeted containment
- Preserve sensor, model, action and operator traces while engaging safety response
Explanation
Correct Answer
D:Correct. Preserve sensor, model, action and operator traces while engaging safety response is aimed at the actual failure: safety incident evidence and physical consequence monitoring are absent. The audit should subsequently obtain unified event reconstruction and emergency-stop telemetry to test execution, not merely the existence of a policy.
Incorrect Answers
A:“Trigger circuit breaker, approved fallback and incident classification based on customer harm” is a control for aI incident runbooks do not define fallback and business-continuity response, but the audit finding here is safety incident evidence and physical consequence monitoring are absent. The relevant measure is “Preserve sensor, model, action and operator traces while engaging safety response”.
B:“Revoke agent tools, halt transactions and invoke financial incident coordination” is a control for emergency controls lack a tested stop mechanism for automated action, but the audit finding here is safety incident evidence and physical consequence monitoring are absent. The relevant measure is “Preserve sensor, model, action and operator traces while engaging safety response”.
C:“Link model-provider identifiers to affected services and activate targeted containment” is a control for supplier incident process lacks AI service dependency mapping, but the audit finding here is safety incident evidence and physical consequence monitoring are absent. The relevant measure is “Preserve sensor, model, action and operator traces while engaging safety response”.
Question 392
In an audit of Incident Response Management Specific to AI, managers say this measure is live: Preserve sensor, model, action and operator traces while engaging safety response. 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?
- Time-stamped failover exercise and incident response timeline
- Tool shutdown telemetry, transaction reconciliation and recovery approval
- Unified event reconstruction and emergency-stop telemetry
- Dependency registry, scoped incident inventory and vendor notice response
Explanation
Correct Answer
C:Correct. The proposed records—unified event reconstruction and emergency-stop telemetry—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is safety incident evidence and physical consequence monitoring are absent.
Incorrect Answers
A:“Time-stamped failover exercise and incident response timeline” would test aI incident runbooks do not define fallback and business-continuity response, not the present safeguard against safety incident evidence and physical consequence monitoring are absent. The expected audit evidence is “Unified event reconstruction and emergency-stop telemetry”.
B:“Tool shutdown telemetry, transaction reconciliation and recovery approval” would test emergency controls lack a tested stop mechanism for automated action, not the present safeguard against safety incident evidence and physical consequence monitoring are absent. The expected audit evidence is “Unified event reconstruction and emergency-stop telemetry”.
D:“Dependency registry, scoped incident inventory and vendor notice response” would test supplier incident process lacks AI service dependency mapping, not the present safeguard against safety incident evidence and physical consequence monitoring are absent. The expected audit evidence is “Unified event reconstruction and emergency-stop telemetry”.
Question 393
In a review of Incident Response Management Specific to AI, investigators cannot establish the expected control proof: unified event reconstruction and emergency-stop telemetry. The underlying event was that an adversarial input creates unsafe physical robot commands and incident records capture only API errors. Which control weakness is the MOST likely cause requiring focused audit work?
- AI incident runbooks do not define fallback and business-continuity response
- Safety incident evidence and physical consequence monitoring are absent
- Emergency controls lack a tested stop mechanism for automated action
- Supplier incident process lacks AI service dependency mapping
Explanation
Correct Answer
B:Correct. The observed condition specifically indicates safety incident evidence and physical consequence monitoring are absent. A meaningful audit response is to verify this diagnosis and then recommend preserve sensor, model, action and operator traces while engaging safety response.
Incorrect Answers
A:“AI incident runbooks do not define fallback and business-continuity response” concerns a separate audit weakness, not the missing assurance described by safety incident evidence and physical consequence monitoring are absent. The correct cause is “Safety incident evidence and physical consequence monitoring are absent”.
C:“Emergency controls lack a tested stop mechanism for automated action” concerns a separate audit weakness, not the missing assurance described by safety incident evidence and physical consequence monitoring are absent. The correct cause is “Safety incident evidence and physical consequence monitoring are absent”.
D:“Supplier incident process lacks AI service dependency mapping” concerns a separate audit weakness, not the missing assurance described by safety incident evidence and physical consequence monitoring are absent. The correct cause is “Safety incident evidence and physical consequence monitoring are absent”.
Question 394
A model incident is closed when metrics recover but the same distribution shift returns each quarter. 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?
- Track trigger conditions, corrective control, owner and verification after closure
- Disable affected retrieval/tool routes, preserve evidence and assess disclosure scope
- Route severity-based AI alerts to accountable on-call incident responders
- Restore manual review controls or pause consequential decisions during degraded mode
Explanation
Correct Answer
A:Correct. Track trigger conditions, corrective control, owner and verification after closure is aimed at the actual failure: root-cause analysis and recurrence prevention are absent from ai incident management. The audit should subsequently obtain recurrent-event analysis and sustained effectiveness checks over later cycles to test execution, not merely the existence of a policy.
Incorrect Answers
B:“Disable affected retrieval/tool routes, preserve evidence and assess disclosure scope” is a control for aI data-exposure playbook lacks containment of model, retrieval and tool paths, but the audit finding here is root-cause analysis and recurrence prevention are absent from ai incident management. The relevant measure is “Track trigger conditions, corrective control, owner and verification after closure”.
C:“Route severity-based AI alerts to accountable on-call incident responders” is a control for aI operational alerts are not integrated into incident severity and on-call ownership, but the audit finding here is root-cause analysis and recurrence prevention are absent from ai incident management. The relevant measure is “Track trigger conditions, corrective control, owner and verification after closure”.
D:“Restore manual review controls or pause consequential decisions during degraded mode” is a control for business-continuity mode disables critical decision safeguards, but the audit finding here is root-cause analysis and recurrence prevention are absent from ai incident management. The relevant measure is “Track trigger conditions, corrective control, owner and verification after closure”.
Question 395
In an audit of Incident Response Management Specific to AI, managers say this measure is live: Track trigger conditions, corrective control, owner and verification after closure. 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?
- Containment timeline, query logs and privacy incident assessment
- Pager traces and tested acknowledgement/resolution targets
- Fail-safe drill results and documented manual approval records
- Recurrent-event analysis and sustained effectiveness checks over later cycles
Explanation
Correct Answer
D:Correct. The proposed records—recurrent-event analysis and sustained effectiveness checks over later cycles—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is root-cause analysis and recurrence prevention are absent from ai incident management.
Incorrect Answers
A:“Containment timeline, query logs and privacy incident assessment” would test aI data-exposure playbook lacks containment of model, retrieval and tool paths, not the present safeguard against root-cause analysis and recurrence prevention are absent from ai incident management. The expected audit evidence is “Recurrent-event analysis and sustained effectiveness checks over later cycles”.
B:“Pager traces and tested acknowledgement/resolution targets” would test aI operational alerts are not integrated into incident severity and on-call ownership, not the present safeguard against root-cause analysis and recurrence prevention are absent from ai incident management. The expected audit evidence is “Recurrent-event analysis and sustained effectiveness checks over later cycles”.
C:“Fail-safe drill results and documented manual approval records” would test business-continuity mode disables critical decision safeguards, not the present safeguard against root-cause analysis and recurrence prevention are absent from ai incident management. The expected audit evidence is “Recurrent-event analysis and sustained effectiveness checks over later cycles”.
Question 396
An internal auditor plans to assess a lending model but has only the application name and no data-flow inventory. 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?
- Plan corroborating tests and independent data extraction for material risk assertions
- Map each outsourced AI component, assurance rights and retained responsibilities
- Identify models, datasets, interfaces, third parties and human decision points before scoping
- Establish audit authority for AI evidence, experts and escalation to governance
Explanation
Correct Answer
C:Correct. Identify models, datasets, interfaces, third parties and human decision points before scoping is aimed at the actual failure: audit scope lacks a complete ai asset and dependency boundary. The audit should subsequently obtain verified architecture and asset inventory reconciled to production usage to test execution, not merely the existence of a policy.
Incorrect Answers
A:“Plan corroborating tests and independent data extraction for material risk assertions” is a control for independence and reliability of management evidence were not assessed, but the audit finding here is audit scope lacks a complete AI asset and dependency boundary. The relevant measure is “Identify models, datasets, interfaces, third parties and human decision points before scoping”.
B:“Map each outsourced AI component, assurance rights and retained responsibilities” is a control for third-party boundaries and shared control responsibility are outside audit scope, but the audit finding here is audit scope lacks a complete AI asset and dependency boundary. The relevant measure is “Identify models, datasets, interfaces, third parties and human decision points before scoping”.
D:“Establish audit authority for AI evidence, experts and escalation to governance” is a control for aI audit mandate and decision-rights access are not explicit, but the audit finding here is audit scope lacks a complete AI asset and dependency boundary. The relevant measure is “Identify models, datasets, interfaces, third parties and human decision points before scoping”.
Question 397
In an audit of Audit Planning and Design, managers say this measure is live: Identify models, datasets, interfaces, third parties and human decision points before scoping. 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?
- Approved audit program specifying independent corroboration and source integrity
- Verified architecture and asset inventory reconciled to production usage
- Contract map and audit scope covering provider and internal control interfaces
- Charter or engagement approval and agreed unrestricted evidence access
Explanation
Correct Answer
B:Correct. The proposed records—verified architecture and asset inventory reconciled to production usage—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is audit scope lacks a complete ai asset and dependency boundary.
Incorrect Answers
A:“Approved audit program specifying independent corroboration and source integrity” would test independence and reliability of management evidence were not assessed, not the present safeguard against audit scope lacks a complete AI asset and dependency boundary. The expected audit evidence is “Verified architecture and asset inventory reconciled to production usage”.
C:“Contract map and audit scope covering provider and internal control interfaces” would test third-party boundaries and shared control responsibility are outside audit scope, not the present safeguard against audit scope lacks a complete AI asset and dependency boundary. The expected audit evidence is “Verified architecture and asset inventory reconciled to production usage”.
D:“Charter or engagement approval and agreed unrestricted evidence access” would test aI audit mandate and decision-rights access are not explicit, not the present safeguard against audit scope lacks a complete AI asset and dependency boundary. The expected audit evidence is “Verified architecture and asset inventory reconciled to production usage”.
Question 398
In a review of Audit Planning and Design, investigators cannot establish the expected control proof: verified architecture and asset inventory reconciled to production usage. The underlying event was that an internal auditor plans to assess a lending model but has only the application name and no data-flow inventory. Which control weakness is the MOST likely cause requiring focused audit work?
- Audit scope lacks a complete AI asset and dependency boundary
- Independence and reliability of management evidence were not assessed
- Third-party boundaries and shared control responsibility are outside audit scope
- AI audit mandate and decision-rights access are not explicit
Explanation
Correct Answer
A:Correct. The observed condition specifically indicates audit scope lacks a complete ai asset and dependency boundary. A meaningful audit response is to verify this diagnosis and then recommend identify models, datasets, interfaces, third parties and human decision points before scoping.
Incorrect Answers
B:“Independence and reliability of management evidence were not assessed” concerns a separate audit weakness, not the missing assurance described by audit scope lacks a complete AI asset and dependency boundary. The correct cause is “Audit scope lacks a complete AI asset and dependency boundary”.
C:“Third-party boundaries and shared control responsibility are outside audit scope” concerns a separate audit weakness, not the missing assurance described by audit scope lacks a complete AI asset and dependency boundary. The correct cause is “Audit scope lacks a complete AI asset and dependency boundary”.
D:“AI audit mandate and decision-rights access are not explicit” concerns a separate audit weakness, not the missing assurance described by audit scope lacks a complete AI asset and dependency boundary. The correct cause is “Audit scope lacks a complete AI asset and dependency boundary”.
Question 399
An audit plan gives the same test effort to a low-stakes translation tool and a clinical triage engine. 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?
- Map each outsourced AI component, assurance rights and retained responsibilities
- Reassess emerging use cases and update the audit universe each planning cycle
- Update engagement objectives for the incident, impact pathways and corrective controls
- Prioritize scope using impact, autonomy, exposure and control maturity
Explanation
Correct Answer
D:Correct. Prioritize scope using impact, autonomy, exposure and control maturity is aimed at the actual failure: risk assessment does not determine assurance depth by potential harm. The audit should subsequently obtain documented risk-based audit universe and engagement resource allocation to test execution, not merely the existence of a policy.
Incorrect Answers
A:“Map each outsourced AI component, assurance rights and retained responsibilities” is a control for third-party boundaries and shared control responsibility are outside audit scope, but the audit finding here is risk assessment does not determine assurance depth by potential harm. The relevant measure is “Prioritize scope using impact, autonomy, exposure and control maturity”.
B:“Reassess emerging use cases and update the audit universe each planning cycle” is a control for new intended uses and affected persons are omitted from risk-based audit planning, but the audit finding here is risk assessment does not determine assurance depth by potential harm. The relevant measure is “Prioritize scope using impact, autonomy, exposure and control maturity”.
C:“Update engagement objectives for the incident, impact pathways and corrective controls” is a control for dynamic risks do not trigger audit plan reconsideration, but the audit finding here is risk assessment does not determine assurance depth by potential harm. The relevant measure is “Prioritize scope using impact, autonomy, exposure and control maturity”.
Question 400
In an audit of Audit Planning and Design, managers say this measure is live: Prioritize scope using impact, autonomy, exposure and control maturity. 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?
- Contract map and audit scope covering provider and internal control interfaces
- Change inventory and documented revisions to annual audit priorities
- Documented risk-based audit universe and engagement resource allocation
- Revised risk assessment and approved audit scope change
Explanation
Correct Answer
C:Correct. The proposed records—documented risk-based audit universe and engagement resource allocation—are directly tied to the asserted safeguard and allow its actual operation to be corroborated. The gap being tested is risk assessment does not determine assurance depth by potential harm.
Incorrect Answers
A:“Contract map and audit scope covering provider and internal control interfaces” would test third-party boundaries and shared control responsibility are outside audit scope, not the present safeguard against risk assessment does not determine assurance depth by potential harm. The expected audit evidence is “Documented risk-based audit universe and engagement resource allocation”.
B:“Change inventory and documented revisions to annual audit priorities” would test new intended uses and affected persons are omitted from risk-based audit planning, not the present safeguard against risk assessment does not determine assurance depth by potential harm. The expected audit evidence is “Documented risk-based audit universe and engagement resource allocation”.
D:“Revised risk assessment and approved audit scope change” would test dynamic risks do not trigger audit plan reconsideration, not the present safeguard against risk assessment does not determine assurance depth by potential harm. The expected audit evidence is “Documented risk-based audit universe and engagement resource allocation”.