引導式測驗 · 中等 等級

SOX 控制中的人工智慧測試測驗

測試演練與 AI 起草的敘述有何不同,為什麼 SoD 分析使用權限級別數據,以及報告和 AI 運行的控制項如何需要自己的證據。

問題 1 的 8

A tool produces a polished purchase-to-pay narrative from a recorded interview. Why is that not a completed walkthrough under AS 2201?

預覽每個問題8 問題

當您參加測驗時,就會出現答案和解釋。

  1. A tool produces a polished purchase-to-pay narrative from a recorded interview. Why is that not a completed walkthrough under AS 2201?
    • AS 2201 prohibits recording interviews
    • A walkthrough follows a transaction and combines inquiry with observation, inspection or re-performance
    • Narratives must be written by the engagement partner
    • Only management can document walkthroughs
  2. In a well-designed evidence-review pipeline, what role should the language model play?
    • Deciding whether each control passes or fails
    • Replacing the IT general controls over the ERP
    • Extracting structured fields, with deterministic rules applying the test
    • Setting sample sizes for the auditor
  3. Why should a segregation-of-duties analysis in SAP look at authorization objects rather than role names?
    • Role names can hide powerful access that only the underlying permissions reveal
    • Role names are encrypted and cannot be read
    • Auditors are forbidden from reviewing role names
    • Authorization objects change every day
  4. Management uses an AI model to perform a monthly reconciliation review control. What does that model now need?
    • Nothing, because AI tools are outside ICFR
    • Approval from the SEC
    • A new COSO component
    • IT general controls such as change management over the model and its configuration
  5. A test uses a system-generated list of all vendor master changes. What must be established about that list?
    • That it was exported by the CFO
    • That it is complete and accurate
    • That it has fewer than 1,000 rows
    • That it was produced by an AI tool
  6. What condition allows benchmarking of an automated application control under AS 2201?
    • Effective IT general controls and evidence that the control has not changed
    • A vendor certificate stating the control works
    • Testing the control only in the fourth quarter
    • Using AI to re-perform the control every day
  7. An analysis finds users who can both create vendors and release payments. What is the usual next step?
    • Automatically delete the users' accounts
    • Report every conflict as a material weakness
    • Ignore the conflicts if the users are managers
    • Check whether mitigating controls cover each conflict and whether the access is needed
  8. Which statement about automation in SOX testing does the guide present as a misconception?
    • Automation can widen coverage from a sample to the whole population
    • Low-confidence extractions should go to human review
    • Automating a test makes the underlying control effective
    • Emergency access logs need review