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
Q631 A company wants to ensure that every AI deployment is reviewed for security before launch. Which mechanism is the BEST?
- A. A review performed only after launch
- B. A voluntary checklist teams may skip
- C. A mandatory pre-deployment security and risk review gate in the delivery process
- D. A review performed by the team that built the system alone
Answer: C
A mandatory gate with independent review ensures consistent security checks before launch. Voluntary, post-launch or self-review mechanisms are unreliable.
Domain 3: AI Technologies and Controls
Q632 A board asks how often AI risk should be reported to it. Which approach is the MOST appropriate?
- A. Reporting only once a year regardless of events
- B. Never, because AI risk is technical
- C. Regular reporting on agreed metrics, with immediate escalation of significant incidents
- D. Only when a regulator requests it
Answer: C
Regular reporting with escalation keeps the board informed of changing risk. Annual or on-request reporting is too infrequent for fast-moving AI risk.
Domain 1: AI Governance and Program Management
Q633 A security team wants to reuse its existing IT standard operating procedures for a new machine learning project. Which gap should the team address FIRST?
- A. Defer procedure updates until best practices for AI procedures are fully settled
- B. Limit the procedures to the IT department because other units do not use AI tools
- C. Confirm that the existing IT procedures already cover every AI use case, and revisit them only after a failed audit
- D. Add procedures for collecting, handling and protecting the data used to train and validate models
Answer: D
AI solutions depend on large volumes of data, so SOPs must add data handling for training and validation. AI is not just another IT system, and procedures should extend across the enterprise and align with the AI acceptable use policy.
Domain 1: AI Governance and Program Management
Q634 An enterprise is drafting AI procedures and manuals for all business units. Which alignment is MOST important?
- A. Procedures should focus on model accuracy and leave ethical concerns to developers
- B. Procedures should align with the AI acceptable use policy and responsible AI principles
- C. Procedures should copy the vendor's default documentation without enterprise review
- D. Procedures should be written separately by each unit so they can differ in risk tolerance
Answer: B
The manual says procedures should align with the AI acceptable use policy and responsible AI concepts for consistent, trustworthy AI. Unit-by-unit divergence, unreviewed vendor text and ignoring ethics all weaken consistency.
Domain 1: AI Governance and Program Management
Q635 A manager asks a generative AI tool to write a network scanning procedure and plans to publish it as written. What is the BEST response from the security lead?
- A. Treat the output as a draft and have it reviewed and verified through standard practices before release
- B. Let the tool approve its own output after a second prompt confirms the content
- C. Reject the tool entirely because AI cannot support documentation work of any kind
- D. Publish it immediately because generative tools follow industry standards by design and the manager owns the request
Answer: A
Generative output needs human oversight and is a starting point. The draft should be circulated for review and verified. Publishing unreviewed text or banning the tool outright both miss the balanced approach.
Domain 1: AI Governance and Program Management
Q636 Before using a generative AI tool to produce procedures, a team defines which processes need procedures and the scope of the request. What is this step MOST likely to improve?
- A. Prompts can be tailored and the output can be validated before it is communicated
- B. The tool will stop producing text that resembles existing documents, which removes any need to review for copying
- C. The model will no longer need any human review of its answers
- D. Training data for the model is automatically deleted afterward
Answer: A
Defining the scope or ask helps tailor prompts and makes output easier to validate. It does not remove the need for human oversight or change how the underlying model is trained.
Domain 1: AI Governance and Program Management
Q637 A company wants to use AI to keep its procedures current over time. Which use is MOST consistent with the manual?
- A. Let the AI rewrite approved procedures overnight without any change record
- B. Replace all procedure owners with a language model that answers staff questions and applies changes on its own
- C. Analyze user feedback and sentiment and flag bottlenecks so procedures can be improved
- D. Remove user feedback channels because the model already knows what users need
Answer: C
NLP and analytics can analyze feedback and sentiment and highlight anomalies or bottlenecks for improvement. Unsupervised rewrites and removing owners or feedback undermine governance.
Domain 1: AI Governance and Program Management
Q638 Which statement about natural language processing in the creation of procedures is MOST accurate?
- A. It removes the need to define which processes require procedures, since the tool infers scope from the documents alone
- B. It guarantees that every generated procedure is free of factual errors
- C. It works best when the source documents are left undefined
- D. It can analyze existing process documents and produce consistent formatting
Answer: D
NLP and natural language generation can analyze existing documentation to create consistent SOP formatting and extract workflows and keywords. Output still needs scoping and human review.
Domain 1: AI Governance and Program Management
Q639 A product team is about to start a new AI project. When should ethical considerations be reviewed to be MOST effective?
- A. During final testing, once the design and training data are locked in place
- B. After launch, once complaints from users have been collected, categorized and escalated to a review board
- C. Early in development, as part of risk assessment and the change management process
- D. Only when a regulator asks for evidence of ethical review
Answer: C
The manual says ethical considerations should be reviewed early, integrated with risk assessment and change management, and prior to project initiation. Reviewing after launch or only on request lets harm occur first.
Domain 1: AI Governance and Program Management
Q640 An organization wants a structured way to evaluate the design, development and deployment of a new AI system before and after release. Which approach BEST fits?
- A. An annual survey of employee opinions about the vendor's brand
- B. A financial audit of the AI project budget after the system goes live
- C. An ethical impact assessment for each new AI solution and change
- D. A penetration test of the production network performed once per year
Answer: C
UNESCO defines an ethical impact assessment as evaluating design, development and deployment and assessing risk before and after release. The manual says one should be performed for new solutions and changes.