Which is an example of an AI-specific risk for the risk register?
Confident but wrong outputs, or hallucinations, are AI-specific risks.
Question 2 of 10 · AI Operations
Why should AI system logs be retained?
Logs provide evidence for investigation and oversight, subject to privacy limits.
Question 3 of 10 · AI Auditing Tools and Techniques
What should an AI audit report include about limitations?
The report should disclose any scope limitations and areas not tested, such as a vendor model that could not be examined, so readers understand the basis and limits of the conclusions. Reporting only positive findings breaches objectivity, a blanket statement of completeness would mislead, and disclosure does not depend on whether management asks.
Question 4 of 10 · AI Governance and Risk
Which governance body should approve high-risk AI uses?
High-risk uses need approval at a defined level of authority.
Question 5 of 10 · AI Operations
An auditor finds that a model's performance has declined since deployment. What is the MOST likely cause?
Changes in the input distribution degrade performance over time.
Question 6 of 10 · AI Auditing Tools and Techniques
What should an auditor document about prompts used with a generative AI audit tool?
Work papers should record the prompts used, the tool and model versions, and how outputs were checked, so a reviewer can understand and reperform the work. The final answer alone, a summary written afterward, or the tool's name and date do not show how the result was produced or validated.
Question 7 of 10 · AI Governance and Risk
An auditor is planning a review of AI governance. What should the auditor obtain FIRST?
The inventory defines the population and owners; without it the auditor cannot scope or sample.
Question 8 of 10 · AI Operations
What is the MAIN purpose of monitoring a deployed model?
Monitoring catches performance and behavior changes that emerge in production.
Question 9 of 10 · AI Auditing Tools and Techniques
Which sampling approach is BEST for testing AI decisions for errors?
Auditor-selected samples that represent the population avoid bias.
Question 10 of 10 · AI Governance and Risk
What should an organization do when an AI vendor changes its terms to allow training on customer data?
A vendor's right to train on customer data changes the confidentiality, privacy and intellectual property risk of the service, so the organization should assess the change against its policies and obligations, then renegotiate, opt out or restrict what data is used. Accepting it without assessment ignores the risk, terminating before any assessment may be disproportionate, and waiting for renewal leaves data exposed in the meantime.
0 of 10
AI Governance and Risk
AI Operations
AI Auditing Tools and Techniques
Which domains cost you the points? Members see a breakdown by domain and a study plan built from it.