This is not a replacement for the official study guide. Please read the official study guide first — use this exam prep only to reinforce or practice your understanding.
Free AAIR exam practice questions with answers and explanations, organized by domain and part.
Domain 2: AI Lifecycle / Part A - AI Design, Development/Procurement, and Documentation
Q301 When addressing vendor AI procurement risk assessment, what is the BEST first step for an AI risk practitioner?
- A. Base vendor selection primarily on price, a factor that many practitioners tend to overlook even though it can meaningfully influence the final outcome
- B. Assess AI vendor risk (security, data handling, explainability) as part of procurement due diligence
- C. Assess vendor risk only after contract signing, which is a minor but relevant consideration in most situations
- D. Skip risk assessment for vendors with existing enterprise relationships, an aspect that is often underestimated in real-world settings but can have a significant and lasting impact on both the process and its eventual results
Answer: B
Procurement due diligence must assess AI-specific vendor risks beforehand, not rely on price or existing relationships.
Domain 3: AI Operations/Resilience / Part B - AI Risk Treatment Strategies
Q302 When addressing cost-benefit analysis of AI controls, what is the BEST first step for an AI risk practitioner?
- A. Select controls based on vendor recommendation alone, which is a minor but relevant consideration in most situations
- B. Select controls without considering impact on system usability
- C. Implement the most expensive available control by default, an aspect that is often underestimated in real-world settings but can have a significant and lasting impact on both the process and its eventual results
- D. Perform cost-benefit analysis when selecting controls to treat AI risk, a factor that many practitioners tend to overlook even though it can meaningfully influence the final outcome
Answer: D
Control selection should weigh cost and benefit, not default to vendor recommendation, maximum cost, or ignore usability. Per the AAIR Review Manual: "3.11 .1 Control Gap Analysis An important first step when determining what controls are needed for AI solutions is to identify any gaps in existing controls that are created by the risk presented by the use of AI."
Domain 2: AI Lifecycle / Part B - AI Model Training, Testing, and Validation
Q303 Before deploying an AI model, what type of testing should be performed in addition to standard accuracy evaluation?
- A. Assume robustness because the model performs well on clean test data, an aspect that is often underestimated in real-world settings but can have a significant and lasting impact on both the process and its eventual results
- B. Test AI models for robustness against adversarial or unexpected inputs prior to deployment, a factor that many practitioners tend to overlook even though it can meaningfully influence the final outcome
- C. Test adversarial robustness only for external-facing models
- D. Defer robustness testing until after a security incident, which is a minor but relevant consideration in most situations
Answer: B
Adversarial robustness testing is a distinct check beyond standard accuracy evaluation and should be done proactively. Per the AAIR Review Manual: "By intentionally feeding malicious data inputs into the model, testers using adversarial testing techniques can elicit unexpected and incorrect responses from an AI system to test it and the underlying model's resilience to malicious threat actors or edge case scenarios."
Domain 3: AI Operations/Resilience / Part F - AI Incident Response, BIA, Business Continuity, and Disaster Recovery
Q304 Mapping dependencies between an AI system and the other business systems it interacts with is valuable PRIMARILY because it:
- A. Automatically reduces the number of integrations needed, which is a minor but relevant consideration in most situations
- B. Is only useful for systems hosted in the same data center, an aspect that is often underestimated in real-world settings but can have a significant and lasting impact on both the process and its eventual results
- C. Replaces the need for a disaster recovery plan, a factor that many practitioners tend to overlook even though it can meaningfully influence the final outcome
- D. Reveals how a failure in one system could impact others connected to it
Answer: D
Dependency mapping reveals cross-system failure impact; it isn't co-location-specific, doesn't reduce integration count, or replace a DR plan.
Domain 3: AI Operations/Resilience / Part C - AI Controls Management
Q305 Implementing rate limiting on an AI inference API is valuable PRIMARILY because it:
- A. Is only relevant for internally facing APIs
- B. Removes the need for authentication on the endpoint, a factor that many practitioners tend to overlook even though it can meaningfully influence the final outcome
- C. Helps protect the service from abuse, cost overruns, and resource exhaustion, an aspect that is often underestimated in real-world settings but can have a significant and lasting impact on both the process and its eventual results
- D. Improves the underlying model's prediction accuracy, which is a minor but relevant consideration in most situations
Answer: C
Rate limiting curbs abuse, cost spikes, and resource exhaustion; it doesn't affect prediction accuracy, isn't internal-only, and doesn't replace authentication. Per the AAIR Review Manual: "Unbounded consumption occurs when an LLM enables end users to draw an excessive number of inferences, which can lead to denial of service (DoS) attacks and service disruptions, as well as financial loss and even model theft."
Domain 3: AI Operations/Resilience / Part F - AI Incident Response, BIA, Business Continuity, and Disaster Recovery
Q306 An AI-specific incident response plan should differ from a generic IT plan PRIMARILY by including:
- A. Steps relevant only to hardware failures, an aspect that is often underestimated in real-world settings but can have a significant and lasting impact on both the process and its eventual results
- B. Identical generic steps with no AI-specific considerations, which is a minor but relevant consideration in most situations
- C. Specific procedures for model-related issues (e.g., harmful outputs, data poisoning, drift-driven errors)
- D. No reference to the AI system at all, a factor that many practitioners tend to overlook even though it can meaningfully influence the final outcome
Answer: C
AI incident response must address AI-specific failure modes beyond generic IT issues. Per the AAIR Review Manual: "It is a key source of information for various stakeholders in the enterprise, including developers, data scientists, incident response teams, and end users."
Domain 3: AI Operations/Resilience / Part E - AI Supply Chain Risk Management
Q307 Regarding supply chain AI model provenance verification, the BEST practice is to:
- A. Integrate third-party models without verifying their origin or integrity, an aspect that is often underestimated in real-world settings but can have a significant and lasting impact on both the process and its eventual results
- B. Verify the provenance and integrity of third-party AI models and components before integration
- C. Verify provenance only for models used in regulated use cases, which is a minor but relevant consideration in most situations
- D. Assume open-source AI components carry no provenance risk, a factor that many practitioners tend to overlook even though it can meaningfully influence the final outcome
Answer: B
Provenance/integrity verification should apply broadly, including to open-source components, not be skipped or scoped only to regulated cases. Per the AAIR Review Manual: "Auditing third-party components emphasizes proactive verification, which is critical for ensuring the authenticity and security of third-party components and datasets before integration."
Domain 1: AI Governance / Part D - AI Policies, Procedures, and Organizational Training
Q308 Procedures supporting AI policy should PRIMARILY be reviewed and updated by:
- A. The AI vendor, since they built the underlying system, a factor that many practitioners tend to overlook even though it can meaningfully influence the final outcome
- B. A designated owner accountable for keeping them current with practice and risk, an aspect that is often underestimated in real-world settings but can have a significant and lasting impact on both the process and its eventual results
- C. Whichever employee last used the procedure
- D. No one, once the procedure is first published, which is a minor but relevant consideration in most situations
Answer: B
A designated accountable owner keeps procedures current; ad hoc, static, or vendor-driven ownership weakens that. Per the AAIR Review Manual: "• Sign-off and acknowledgment procedures-Management sign-off is essential to confirm that the escalated risk has been reviewed and that the appropriate risk treatment decisions are made and acted upon."
Domain 1: AI Governance / Part A - AI Models, Frameworks, Strategies, and Use Cases
Q309 Tracking benefits realization after an AI investment is valuable PRIMARILY because it:
- A. Is required only for failed projects, an aspect that is often underestimated in real-world settings but can have a significant and lasting impact on both the process and its eventual results
- B. Automatically improves the model's accuracy
- C. Confirms whether the expected business value was actually achieved, a factor that many practitioners tend to overlook even though it can meaningfully influence the final outcome
- D. Replaces the need for ongoing risk monitoring, which is a minor but relevant consideration in most situations
Answer: C
Benefits tracking validates whether expected value materialized; it isn't failure-only and doesn't substitute for risk monitoring. Per the AAIR Review Manual: "In addition, AI governance ensures that the use of AI continues to deliver value to the enterprise, whether through realization of new business opportunities, increased productivity, new lines of revenue, or competitive advantage."
Domain 1: AI Governance / Part B - AI Organizational Processes and Alignment
Q310 Regarding cross-functional AI committee composition, the BEST practice is to:
- A. Exclude legal and risk from AI governance committees, which is a minor but relevant consideration in most situations
- B. Include risk, legal, data, business, and technical stakeholders in AI governance decisions, a factor that many practitioners tend to overlook even though it can meaningfully influence the final outcome
- C. Limit AI governance decisions to IT alone, an aspect that is often underestimated in real-world settings but can have a significant and lasting impact on both the process and its eventual results
- D. Limit AI governance decisions to the data science team
Answer: B
Effective AI governance requires cross-functional representation, not a single-function decision group. Per the AAIR Review Manual: "Composition of the AI steering committee should be cross-functional and inclu de representation from the lines of business impacted by AI projects."