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 1: AI Governance and Program Management
Q511 A leadership team wants to ensure that AI output is accountable and traceable. Which program element is MOST relevant?
- A. Outsourcing accountability for AI decisions to the vendor
- B. Maximizing model size so that every decision is better justified
- C. Reducing logging so that audit evidence is easier to store
- D. Mandating audits and traceability for AI systems
Answer: D
Mandating audits and traceability supports accountability for AI decisions. Model size and reduced logging do not improve accountability, and outsourcing cannot transfer it.
Domain 1: AI Governance and Program Management
Q512 The enterprise wants to prepare for workforce effects of AI. Which security program step BEST addresses this?
- A. Eliminating all human roles that AI could perform in the near term, to capture savings as soon as possible
- B. Ignoring workforce concerns so that deployment is not delayed
- C. Banning AI use to avoid any workforce disruption
- D. Societal adaptation through reskilling and communication about changed roles
Answer: D
The program should consider societal adaptation, including reskilling, upskilling and revised roles. Ignoring concerns or eliminating roles damages support, and banning AI forgoes benefits.
Domain 1: AI Governance and Program Management
Q513 Which statement BEST describes aligning the AI security program with organizational objectives?
- A. Security measures should support business goals and be justified by risk and value
- B. Security goals should replace business goals when they conflict
- C. Alignment is only needed for marketing and brand communication
- D. Security should operate independently of the business and its strategy, so that its judgments are not influenced by commercial goals
Answer: A
A program aligned with organizational objectives is justified through risk and value and supports strategy. Independence from the business or replacing goals undermines support and funding.
Domain 1: AI Governance and Program Management
Q514 Leadership asks how to show that AI security investment is delivering results. Which approach is MOST appropriate?
- A. Report only the number of meetings held by the steering group, together with the average attendance at each session
- B. Report only the budget that has been spent on AI security
- C. Define program metrics tied to objectives and report them to stakeholders regularly
- D. Share no results to avoid creating unrealistic expectations
Answer: C
Program metrics tied to objectives and regularly reported to stakeholders show effectiveness. Budget, meetings and silence do not show outcomes.
Domain 1: AI Governance and Program Management
Q515 A security team uses AI to support its decision-making on incident triage. Which risk should it MOST consider?
- A. Overreliance on AI outputs without human review of high-impact decisions
- B. Better detection with no downsides because AI decisions are always correct
- C. Elimination of all false positives from the monitoring tools
- D. No further need for analysts since the AI decides everything
Answer: A
AI can speed triage, but it can produce false positives or errors, so human review is needed for high-impact decisions. AI does not eliminate false positives or analysts.
Domain 1: AI Governance and Program Management
Q516 An organization wants its incident response policy to be AI-ready. Which element should it MOST importantly add?
- A. Only physical security procedures for the data center facilities, such as badge access and visitor logs for the building
- B. Only password rotation rules for the administrators
- C. Only a list of approved vendors and their contact numbers
- D. AI-specific threats, detection and response techniques, and ethical considerations
Answer: D
An AI-ready incident response policy addresses AI-specific threats, detection and response techniques and ethics, plus communication and escalation. Physical, password and vendor content alone does not cover AI incidents.
Domain 1: AI Governance and Program Management
Q517 Which documentation would MOST help incident investigators interpret an AI model's architecture, training data and decision-making?
- A. Lunch menus for the staff attending the investigation
- B. Model cards and related model documentation
- C. Marketing brochures describing the benefits of the AI product
- D. Holiday schedules for the data science team
Answer: B
Thorough documentation such as model cards helps investigators understand the model. Marketing brochures and schedules do not describe architecture or data.
Domain 1: AI Governance and Program Management
Q518 AI makes mistakes by design because it is probabilistic. How should this MOST affect the incident identification process?
- A. Ignore all wrong predictions because errors are normal for AI
- B. Expect zero errors and investigate every deviation from perfection
- C. Define what counts as an AI incident and thresholds for escalating wrong predictions
- D. Treat every wrong prediction as a security breach requiring escalation to the incident commander and the board
Answer: C
Because AI is rooted in probability, organizations need criteria for when errors become incidents. Treating all errors as breaches would overwhelm teams, and ignoring them or expecting zero errors is unrealistic.
Domain 1: AI Governance and Program Management
Q519 During the assess phase, a team determines that an AI incident has exposed personal data from a training set. What should the team MOST appropriately do?
- A. Evaluate impact on individuals and regulatory notification duties with privacy experts
- B. Wait for customers to complain before taking any further action, so that the response is proportionate to the actual demand
- C. Delete the model and all evidence to limit further exposure
- D. Close the incident without review since no systems were damaged
Answer: A
Privacy experts guide response on impact to personal data, data subject rights and compliance obligations. Closing, deleting evidence or waiting for complaints is inadequate.
Domain 1: AI Governance and Program Management
Q520 A model is behaving unsafely and threatens human safety. What should be done FIRST?
- A. Convene a long committee review before acting on the problem
- B. Notify only the marketing team so that customer messaging can be prepared before the technical response begins
- C. Take immediate action to stop the harm, even if containment is complex
- D. Wait for the next scheduled release to correct the behavior
Answer: C
Threats to human life and safety require immediate action. Containment may be hard because of AI complexity but cannot be delayed. Committees, release schedules and marketing notices are too slow or irrelevant.