AAISM Practice Questions
Prepare for AAISM with more than an answer.
- Exam fee
- $599 USD
- Level
- Advanced
- Valid for
- Ongoing with annual CPE requirements
Domains covered on the exam 3
- AI Governance and Program Management31%
- AI Risk Management31%
- AI Technologies and Controls38%
- 1
Select TWO key objectives of the 'Manage' function within the NIST AI Risk Management Framework (AI RMF).
Show answer details
Correct answer: A, B
The 'Manage' function focuses on allocating resources to mapped and measured risks, prioritizing them, and applying treatment (response) plans.
Implementing strategies to balance benefits and risks (treatment) is a core part of the 'Manage' function.
- 2
Which ISO standard specifies requirements for establishing, implementing, maintaining, and continually improving an Artificial Intelligence Management System (AIMS)?
Show answer details
Correct answer: D
ISO/IEC 42001 is the international management system standard specifically for AI (AIMS). It follows the same structure as ISO 27001 but focuses on AI-specific processes and controls.
- 3
What is the primary purpose of 'Model Watermarking' in the context of AI security?
Show answer details
Correct answer: B
Model Watermarking involves embedding a unique, hidden signal (backdoor or pattern) into the model's behavior or weights. If a stolen model is found, the owner can query it with specific trigger inputs to reveal the watermark, proving ownership and detecting IP theft.
- 4
A hospital is training a diagnostic model on patient X-rays. They want to ensure that even if the model is inspected, no single patient's data can be reconstructed. They decide to use Differential Privacy (DP). Which parameter in DP controls the privacy budget, where a smaller value indicates stronger privacy?
Show answer details
Correct answer: C
Epsilon (ε) is the primary parameter in Differential Privacy, representing the privacy loss or budget. A smaller epsilon value means the output probability distributions are more similar regardless of a single individual's presence, thus providing stronger privacy (more noise added).
- 5
When conducting an AI Risk Assessment, 'Automation Bias' is a human-centric risk that must be considered. What does this term refer to?
Show answer details
Correct answer: B
Automation Bias is a cognitive bias where human operators trust automated systems (like AI) too much, failing to detect errors or verify outputs. This is a critical risk in human-in-the-loop systems.
- 6
A multinational financial institution is establishing its AI Governance Framework. The Chief AI Security Officer (CAISO) is defining the responsibilities of the AI Steering Committee. According to ISACA best practices and the COBIT for AI rationale, which of the following is the PRIMARY responsibility of this committee regarding AI risk appetite?
Show answer details
Correct answer: B
The AI Steering Committee acts as a bridge between the Board of Directors and operational management. Its primary role regarding risk is to translate the high-level enterprise risk appetite set by the Board into actionable AI-specific risk tolerance levels that guide project approval and control implementation.
- 7
An organization is adopting the NIST AI Risk Management Framework (AI RMF) to govern its generative AI deployments. During the 'Map' function, the team is struggling to identify potential downstream impacts. Which of the following activities is MOST critical during this phase to ensure comprehensive context establishment?
Show answer details
Correct answer: D
The 'Map' function in NIST AI RMF is focused on establishing context. Characterizing the intended purpose, users, and deployment environment is the foundational step to identifying risks. Without this context, downstream risks cannot be accurately mapped.
- 8
You are designing a secure architecture for a healthcare AI model that processes highly sensitive patient data. To minimize privacy risks during the training phase without exposing raw data to the central server, you decide to implement Federated Learning.
Review the diagram below representing the proposed flow. What critical security control is missing from the aggregation step (Server) to prevent the reconstruction of individual user data from gradient updates?
Show answer details
Correct answer: B
In Federated Learning, sending raw gradients to a server can still leak information about the training data (model inversion/inference attacks). To prevent this at the aggregation step, Secure Aggregation (often using SMPC) or adding Differential Privacy (noise) to the updates is required to ensure the server sees only the aggregate update, not individual contributions.
sequenceDiagram participant Client1 participant Client2 participant Server Client1->>Client1: Train Local Model Client2->>Client2: Train Local Model Client1->>Server: Send Gradient (Encrypted/Noisy) Client2->>Server: Send Gradient (Encrypted/Noisy) Note over Server: Secure Aggregation Required Here Server->>Server: Update Global Model Server-->>Client1: Send New Global Model Server-->>Client2: Send New Global Model - 9
A retail company is updating its Acceptable Use Policy (AUP) to address the use of public Generative AI tools by employees. Which of the following provisions is MOST critical to include to prevent data leakage while maintaining productivity?
Show answer details
Correct answer: D
The most critical risk with public GenAI tools is the training of the provider's models on user data. An AUP must explicitly prohibit entering sensitive data (PII, IP, secrets) into public tools, while allowing use for non-sensitive tasks to balance security and productivity.
- 10
Which of the following activities are considered 'High Risk' under the EU AI Act and would require the most stringent conformity assessments? (Select TWO)
Show answer details
Correct answer: A, D
AI systems evaluating creditworthiness of natural persons or establishing their credit score are High Risk as they determine access to essential private services.
Under the EU AI Act, systems used in employment, workers management, and access to self-employment (e.g., CV screening software) are classified as High Risk due to their potential impact on fundamental rights.
