Isaca AAIA Practice Test 72
Question 710
An independent AI auditor reviews this case: A procurement agent can create vendor payments beyond its approval limit because tool scopes mirror an administrator account. Which TWO complementary elements should the auditor seek: an effective corrective control and the most probative independent evidence? Select TWO answers.
- Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces.
- Inspect reviewer workload, UI, explanations, override log and interviews on actual cases.
- Limit agent authority by least-privilege tools, transaction ceilings, pre-action approval and post-action verification.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
- Design actionable challenge, human reassessment, factual correction and remediation for adverse AI decisions.
Explanation
Correct Answers
A:The reason it is necessary is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone. This evidence directly tests the asserted control for Autonomous agent action-limit supervision: Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces.
C:The reason it is necessary is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone. For this situation: A procurement agent can create vendor payments beyond its approval limit because tool scopes mirror an administrator account.
Incorrect Answers
B:This option concerns Effective human oversight of AI recommendations, which is a related but different AI governance concern. For this situation: A procurement agent can create vendor payments beyond its approval limit because tool scopes mirror an administrator account.
D:This attractive shortcut does not resolve the Autonomous agent action-limit supervision weakness: Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully. For this situation: A procurement agent can create vendor payments beyond its approval limit because tool scopes mirror an administrator account.
E:This option concerns Customer appeal and remedy for automated outcomes, which is a related but different AI governance concern. It does not establish control effectiveness for Autonomous agent action-limit supervision.
Question 711
A governance and assurance review uncovers the following: A procurement agent can create vendor payments beyond its approval limit because tool scopes mirror an administrator account. What evidence should the auditor examine FIRST to substantiate or challenge the control owner’s claim? Select ONE best answer.
- Review appeal turnaround, reviewer authority, data correction process, overturned cases and notices.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
- Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces.
- Inspect live KPIs and KRIs, baselines, incident triggers, ownership and corrective decisions.
Explanation
Correct Answer
C:The reason it is necessary is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone. The corrective action is: Limit agent authority by least-privilege tools, transaction ceilings, pre-action approval and post-action verification.
Incorrect Answers
A:This option concerns Customer appeal and remedy for automated outcomes, which is a related but different AI governance concern. For this situation: A procurement agent can create vendor payments beyond its approval limit because tool scopes mirror an administrator account.
B:This attractive shortcut does not resolve the Autonomous agent action-limit supervision weakness: Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully. The auditor should instead examine: Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces.
D:This option concerns AI post-deployment monitoring and KRI ownership, which is a related but different AI governance concern. It does not establish control effectiveness for Autonomous agent action-limit supervision.
Question 712
A risk-based audit engagement examines this operating condition: A procurement agent can create vendor payments beyond its approval limit because tool scopes mirror an administrator account. After management addresses the issue, which result MOST convincingly demonstrates effective remediation? Select ONE best answer.
- Unauthorized changes are reversed or contained and renewed tool access passes adversarial gate tests.
- Threshold breaches lead to timely investigation and documented changes to protect affected users.
- Unapproved high-impact actions are blocked before execution, and approved actions are attributable.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
Explanation
Correct Answer
C:The reason it is necessary is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone. This is an observable validation result for Autonomous agent action-limit supervision, not just management assurance: Unapproved high-impact actions are blocked before execution, and approved actions are attributable.
Incorrect Answers
A:This option concerns AI agent unauthorized action incident, which is a related but different AI governance concern. For this situation: A procurement agent can create vendor payments beyond its approval limit because tool scopes mirror an administrator account.
B:This option concerns AI post-deployment monitoring and KRI ownership, which is a related but different AI governance concern. For this situation: A procurement agent can create vendor payments beyond its approval limit because tool scopes mirror an administrator account.
D:This attractive shortcut does not resolve the Autonomous agent action-limit supervision weakness: Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully. Effective closure requires: Unapproved high-impact actions are blocked before execution, and approved actions are attributable.
Question 713
An audit committee asks for a defensible assessment of this situation: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward. What is the MOST appropriate risk-based audit recommendation that addresses the root control weakness? Select ONE best answer.
- Verify feedback provenance, reviewer authority and sampling before incorporating live corrections into future models.
- Limit agent authority by least-privilege tools, transaction ceilings, pre-action approval and post-action verification.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
- Disable or restrict affected tools, preserve transaction records and verify downstream changes before re-enabling.
Explanation
Correct Answer
B:The reason it is necessary is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone. In this case: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward.
Incorrect Answers
A:This option concerns Feedback-loop contamination from user corrections, which is a related but different AI governance concern. It does not establish control effectiveness for Autonomous agent action-limit supervision.
C:This attractive shortcut does not resolve the Autonomous agent action-limit supervision weakness: Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully. In this case: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward.
D:This option concerns AI agent unauthorized action incident, which is a related but different AI governance concern. The decisive principle here is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone.
Question 714
During an AI controls assessment, the team learns that an AI operations assistant can delete cloud backups while a human reviewer is notified only afterward. Which principle BEST explains why the observed practice is a material governance or control concern? Select ONE best answer.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
- Human-in-the-loop labeling does not create real oversight when reviewers cannot understand or influence outcomes.
- Thresholds are business control decisions whose error costs vary by context and input quality.
- Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone.
Explanation
Correct Answer
D:Independent audit evidence includes: Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces. In this case: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward.
Incorrect Answers
A:This attractive shortcut does not resolve the Autonomous agent action-limit supervision weakness: Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully. For this situation: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward.
B:This option concerns Effective human oversight of AI recommendations, which is a related but different AI governance concern. The decisive principle here is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone.
C:This option concerns Threshold governance and safety fallbacks, which is a related but different AI governance concern. It does not establish control effectiveness for Autonomous agent action-limit supervision.
Question 715
An independent AI auditor reviews this case: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward. What evidence should the auditor examine FIRST to substantiate or challenge the control owner’s claim? Select ONE best answer.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
- Inspect reviewer workload, UI, explanations, override log and interviews on actual cases.
- Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces.
- Inspect tool invocation audit, actor scopes, approval checks, external transaction state and rollback capability.
Explanation
Correct Answer
C:The reason it is necessary is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone. The acceptable verified outcome is: Unapproved high-impact actions are blocked before execution, and approved actions are attributable.
Incorrect Answers
A:This attractive shortcut does not resolve the Autonomous agent action-limit supervision weakness: Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully. Limit agent authority by least-privilege tools, transaction ceilings, pre-action approval and post-action verification.
B:This option concerns Effective human oversight of AI recommendations, which is a related but different AI governance concern. For this situation: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward.
D:This option concerns AI agent unauthorized action incident, which is a related but different AI governance concern. For this situation: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward.
Question 716
A governance and assurance review uncovers the following: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward. After management addresses the issue, which result MOST convincingly demonstrates effective remediation? Select ONE best answer.
- Uncertain cases route safely and measured errors remain within approved operational and fairness limits.
- Unapproved high-impact actions are blocked before execution, and approved actions are attributable.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
- Approved feedback improves measured outcomes without adversarial self-reinforcing error.
Explanation
Correct Answer
B:This is an observable validation result for Autonomous agent action-limit supervision, not just management assurance: Unapproved high-impact actions are blocked before execution, and approved actions are attributable.
Incorrect Answers
A:This option concerns Threshold governance and safety fallbacks, which is a related but different AI governance concern. For this situation: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward.
C:For this situation: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward.
D:This option concerns Feedback-loop contamination from user corrections, which is a related but different AI governance concern. For this situation: An AI operations assistant can delete cloud backups while a human reviewer is notified only afterward.
Question 717
A risk-based audit engagement examines this operating condition: A customer service agent sends refunds without checking policy ceilings or approved transaction identities. What is the MOST appropriate risk-based audit recommendation that addresses the root control weakness? Select ONE best answer.
- Verify feedback provenance, reviewer authority and sampling before incorporating live corrections into future models.
- Define owner-led ongoing monitoring for quality, drift, safety, fairness and service impact with escalation.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
- Limit agent authority by least-privilege tools, transaction ceilings, pre-action approval and post-action verification.
Explanation
Correct Answer
D:The reason it is necessary is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone. In this case: A customer service agent sends refunds without checking policy ceilings or approved transaction identities.
Incorrect Answers
A:This option concerns Feedback-loop contamination from user corrections, which is a related but different AI governance concern. The decisive principle here is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone.
B:This option concerns AI post-deployment monitoring and KRI ownership, which is a related but different AI governance concern. The decisive principle here is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone.
C:This attractive shortcut does not resolve the Autonomous agent action-limit supervision weakness: Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully. In this case: A customer service agent sends refunds without checking policy ceilings or approved transaction identities.
Question 718
An audit committee asks for a defensible assessment of this situation: A customer service agent sends refunds without checking policy ceilings or approved transaction identities. Which principle BEST explains why the observed practice is a material governance or control concern? Select ONE best answer.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
- Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone.
- Auditability of an AI decision depends on traceable context, not just a final predicted label.
- Production AI monitoring must measure harmful outcomes, not merely infrastructure availability.
Explanation
Correct Answer
B:Independent audit evidence includes: Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces. In this case: A customer service agent sends refunds without checking policy ceilings or approved transaction identities.
Incorrect Answers
A:This attractive shortcut does not resolve the Autonomous agent action-limit supervision weakness: Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully. For this situation: A customer service agent sends refunds without checking policy ceilings or approved transaction identities.
C:This option concerns AI decision logs and explanation traceability, which is a related but different AI governance concern. It does not establish control effectiveness for Autonomous agent action-limit supervision.
D:This option concerns AI post-deployment monitoring and KRI ownership, which is a related but different AI governance concern. For this situation: A customer service agent sends refunds without checking policy ceilings or approved transaction identities.
Question 719
During an AI controls assessment, the team learns that a customer service agent sends refunds without checking policy ceilings or approved transaction identities. What evidence should the auditor examine FIRST to substantiate or challenge the control owner’s claim? Select ONE best answer.
- Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces.
- Inspect live KPIs and KRIs, baselines, incident triggers, ownership and corrective decisions.
- Inspect tool invocation audit, actor scopes, approval checks, external transaction state and rollback capability.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
Explanation
Correct Answer
A:This evidence directly tests the asserted control for Autonomous agent action-limit supervision: Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces.
Incorrect Answers
B:This option concerns AI post-deployment monitoring and KRI ownership, which is a related but different AI governance concern. The audit must corroborate: Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces.
C:This option concerns AI agent unauthorized action incident, which is a related but different AI governance concern. In this case: A customer service agent sends refunds without checking policy ceilings or approved transaction identities.
D:For this situation: A customer service agent sends refunds without checking policy ceilings or approved transaction identities.
Question 720
An independent AI auditor reviews this case: A customer service agent sends refunds without checking policy ceilings or approved transaction identities. Which TWO complementary elements should the auditor seek: an effective corrective control and the most probative independent evidence? Select TWO answers.
- Define owner-led ongoing monitoring for quality, drift, safety, fairness and service impact with escalation.
- Give an AI agent full administrator rights and rely on a natural-language instruction to act carefully.
- Inspect service identities, tool-call logs, policy checks, denial tests and human approval traces.
- Limit agent authority by least-privilege tools, transaction ceilings, pre-action approval and post-action verification.
- Inspect calibration, confusion matrices, missing-input behavior, escalation queue and approval history.
Explanation
Correct Answers
C:The reason it is necessary is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone. For this situation: A customer service agent sends refunds without checking policy ceilings or approved transaction identities.
D:The reason it is necessary is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone. This response fits the risk associated with Autonomous agent action-limit supervision: Limit agent authority by least-privilege tools, transaction ceilings, pre-action approval and post-action verification.
Incorrect Answers
A:This option concerns AI post-deployment monitoring and KRI ownership, which is a related but different AI governance concern. A proportionate response is: Limit agent authority by least-privilege tools, transaction ceilings, pre-action approval and post-action verification.
B:It misses the key control principle: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone.
E:This option concerns Threshold governance and safety fallbacks, which is a related but different AI governance concern. The decisive principle here is: Autonomous action risk is driven by actual tool permissions and decision gates, not chatbot prompt wording alone.
