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 AAISM exam practice questions with answers and explanations, organized by domain and part.
Domain 3: AI Technologies and Controls
Q181 Which of the following is BEST for analyzing true positives, true negatives, false positives, and false negatives produced by an AI model?
- A. Hyperparameter tuning
- B. Precision
- C. Confusion matrix
- D. Recall
Answer: C
A confusion matrix is explicitly defined in AAISM as the framework used to interpret classification performance by listing: • true positives • true negatives • false positives • false negatives Precision (B) and recall (D) are derived metrics that use parts of the matrix but do not show the full picture. Hyperparameter tuning (A) is unrelated.
Domain 2: AI Risk Management
Q182 A SaaS-based LLM system has risks including prompt injection, data poisoning, and model exfiltration. What is the BEST way to ensure consistent risk treatment?
- A. Apply control baselines from a recognized industry standard
- B. Implement an AI threat control matrix mapping threats to controls and assurance
- C. Focus on post-deployment red teaming
- D. Rely on vendor audit reports and SLAs
Answer: B
AAISM highlights the importance of an AI threat–control matrix that: • maps threats (e.g., poisoning, exfiltration, injection) • identifies required controls • defines assurance/testing activities • ensures consistent coverage across all AI components Control baselines (A) are useful but not AI-specific. Red teaming (C) is reactive and not comprehensive. Vendor reports (D) are supplemental, not sufficient.
Domain 1: AI Governance and Program Management
Q183 Which of the following reviews is the MOST essential part of an AI impact assessment?
- A. Testing, evaluation, validation, and verification
- B. Evaluation of model reproducibility
- C. Security control self-assessment (CSA)
- D. Identification of environmental and societal consequences
Answer: D
The AAISM material frames an AI impact assessment as broader than a technical model review. It emphasizes that governance of AI includes assessing societal, environmental, and human-rights impacts in addition to organizational risk. The official guidance notes that high-level impact assessments must explicitly consider “potential consequences for individuals, communities, and the environment” as part of responsible AI governance. Activities such as testing/validation (A) and reproducibility (B) are critical in the development lifecycle but are not, by themselves, sufficient to constitute an impact assessment. A CSA (C) is focused on security controls, which is narrower than overall impact. The mandatory element that differentiates an AI impact assessment is the assessment of environmental and societal consequences, making option D the correct answer in alignment with governance expectations.
Domain 2: AI Risk Management
Q184 During red-team testing of an AI system used to make lending decisions, which of the following techniques BEST simulates a data poisoning attack?
- A. Inputting encrypted data into the model
- B. Adding noise to output predictions
- C. Stealing model weights from a deployed API
- D. Corrupting training data sets to manipulate outcomes
Answer: D
AAISM defines data poisoning as the intentional manipulation of training data so that the learned model behaves incorrectly (e.g., skewed lending approvals/denials) while appearing valid. The correct simulation in red-team exercises is to corrupt or seed the training set with adversarial examples or mislabeled records to induce biased or erroneous decision boundaries. Encrypting inputs (A) is unrelated; output noise (B) describes perturbation of predictions, not training; model weight theft (C) is model extraction, not poisoning.
Domain 3: AI Technologies and Controls
Q185 For a life insurance company deploying AI for fraud detection, which factor is MOST critical?
- A. Robustness
- B. Accuracy
- C. Explainability
- D. Adaptability
Answer: A
AAISM emphasizes robustness as the key requirement for fraud-detection systems because they must resist adversarial manipulation, data poisoning, spoofing, and input tampering. Accuracy (B) matters but does not protect against adversarial attacks. Explainability (C) is important but secondary. Adaptability (D) is useful but not the top security requirement.
Domain 2: AI Risk Management
Q186 After deployment, an AI model’s output begins to drift outside of the expected range. Which of the following is the development team’s BEST course of action?
- A. Take the AI model offline
- B. Adjust the hyperparameters of the AI model
- C. Create an emergency change request to correct the issue
- D. Return to an earlier phase in the AI life cycle
Answer: D
AAISM emphasizes that when model drift occurs, the best response is not a quick fix but rather to revisit an earlier phase of the AI life cycle to address data quality, retraining, or evaluation processes. Simply taking the model offline halts functionality without resolution, while adjusting hyperparameters or issuing emergency changes treats the symptom rather than the root cause. Proper governance requires returning to the design or training phases to re-establish stability, accuracy, and compliance of the model. Thus, the correct approach is to return to an earlier AI lifecycle phase.
Domain 1: AI Governance and Program Management
Q187 A school district contracts a third-party provider for AI-based curriculum recommendations. Which of the following is the BEST way to ensure the vendor uses AI responsibly?
- A. Confirming the AI solution supports single sign-on (SSO)
- B. Verifying the vendor has updated terms of service
- C. Requiring the vendor to provide the model card
- D. Ensuring the vendor offers 24/7 technical support
Answer: C
AAISM emphasizes transparency artifacts from vendors to enable due diligence and assurance. A model card documents intended use, data sources, limitations, performance across subgroups, known risks, and evaluation procedures—information necessary to assess safety, fairness, and compliance for sensitive contexts like education. SSO and support are useful operational features; generic ToS updates are insufficient without model-specific disclosures.
Domain 3: AI Technologies and Controls
Q188 Which of the following controls BEST mitigates the risk of bias in AI models?
- A. Robust access control techniques
- B. Regular data reconciliation
- C. Cryptographic hash functions
- D. Diverse data sourcing strategies
Answer: D
Bias in AI models primarily stems from limitations or imbalances in training data. The AAISM study materials emphasize that the most effective way to mitigate this risk is through diverse data sourcing strategies that ensure coverage across demographics, scenarios, and contexts. Access controls protect data security, not fairness. Data reconciliation ensures accuracy but does not address representational imbalance. Cryptographic hashing preserves integrity but has no impact on bias mitigation. To reduce systemic unfairness, the critical control is sourcing diverse and representative data.
Domain 2: AI Risk Management
Q189 Which attack type is MOST likely to cause model drift?
- A. Model stealing
- B. Perfect knowledge
- C. Data poisoning
- D. Membership inference
Answer: C
AAISM defines data poisoning as directly capable of causing model drift because corrupted training data shifts the statistical distribution, leading to degraded or unsafe performance. Model stealing (A) extracts model behavior but does not cause drift. Perfect knowledge (B) is an attacker capability, not an attack causing drift. Membership inference (D) attacks privacy, not performance.
Domain 2: AI Risk Management
Q190 Which of the following key risk indicators (KRIs) is MOST relevant when evaluating the effectiveness of an organization’s AI risk management program?
- A. Number of AI models deployed into production
- B. Percentage of critical business systems with AI components
- C. Percentage of AI projects in compliance
- D. Number of AI-related training requests submitted
Answer: C
AAISM identifies percentage of AI projects in compliance as the most relevant KRI for evaluating AI risk management effectiveness. This metric directly reflects adherence to governance, regulatory, and security requirements. The number of models deployed (A) or systems with AI components (B) indicate scale, not risk management quality. Training requests (D) show awareness levels but do not measure effectiveness of risk management. Compliance percentage provides a direct, measurable indication of how well risks are being governed and mitigated.