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
Q491 An AI incident is detected. Which action is MOST appropriate FIRST under AI incident management?
- A. Triage and contain using the AI-aware incident response plan
- B. Wait for the vendor to contact the organization about the event
- C. Announce the incident publicly before any analysis is completed
- D. Delete all logs immediately to limit the spread of information
Answer: A
Following the incident response plan, triage and containment come first, using AI-specific playbooks. Deleting logs destroys evidence, and waiting or announcing before analysis is premature.
Domain 3: AI Technologies and Controls
Q492 Which topic is MOST important to include in AI security awareness training for all employees?
- A. Compiler design principles used in deep learning frameworks
- B. Network cable standards used in data center cabling plans
- C. Acceptable use, data handling limits and recognizing AI-enabled phishing or deepfakes
- D. Detailed GPU kernel optimization techniques for training workloads, including memory tiling and mixed precision tuning
Answer: C
Employees need to know acceptable use, what data may be entered into AI tools and how to spot AI-enabled social engineering. Technical optimization topics suit specialists only.
Domain 3: AI Technologies and Controls
Q493 A security team lacks staff who understand AI threats. What is the BEST approach to address this skills gap?
- A. Provide targeted training and certifications and use qualified partners where needed
- B. Delay all AI projects indefinitely until the market matures and the vendors have settled on a single common approach
- C. Ignore AI risk until staff with expertise are available
- D. Hire only marketing staff to promote the AI program
Answer: A
Targeted upskilling and partnering address skills gaps, which the manual lists as a major challenge. Ignoring risk or hiring unrelated staff does not help.
Domain 3: AI Technologies and Controls
Q494 Who is MOST in need of role-specific AI training beyond general awareness?
- A. External caterers who provide meals for corporate events, including those arranged for the AI development teams
- B. Delivery drivers who bring packages to the loading dock
- C. Developers, data scientists and administrators who build or operate AI systems
- D. Visitors to the lobby who enter the building each day
Answer: C
Roles building or operating AI need deeper training on secure development, data handling and model risk. Lobby visitors and caterers have no relevant role.
Domain 3: AI Technologies and Controls
Q495 A fraud model's precision slowly declines as customer behavior changes. Which monitoring activity is MOST appropriate?
- A. Drift detection with retraining or recalibration triggers
- B. Increasing storage capacity for archived prediction logs
- C. Disabling the model permanently so no further errors occur
- D. Changing the dashboard theme so declines are easier to read
Answer: A
Drift monitoring detects data or concept changes and triggers retraining or recalibration. Disabling the model removes protection, and storage or logos do not address drift.
Domain 3: AI Technologies and Controls
Q496 Which monitoring source is MOST useful for staying current on emerging AI threats?
- A. Annual reports published by industry bodies once a year
- B. Rumors passed on by colleagues in informal conversations at industry events and in the corridors of the company
- C. Unmoderated online forums with anonymous threat claims
- D. Threat intelligence from sources such as OWASP, MITRE and vendor advisories
Answer: D
Threat intelligence from reputable sources keeps defenses current. Annual reports are too slow, and rumors or unmoderated forums are unreliable.
Domain 3: AI Technologies and Controls
Q497 Which metric is MOST useful to show AI security program effectiveness to executives?
- A. Trends in incidents, detection and response times, drift events and control coverage
- B. The number of meetings held by the security committee
- C. The average office attendance of the security team
- D. The number of slides produced for the monthly steering meeting and the number of people who attended each presentation
Answer: A
Security metrics should reflect outcomes such as incidents, response times, drift and coverage. Slides, meetings and attendance do not show effectiveness.
Domain 3: AI Technologies and Controls
Q498 An alert shows a spike of unusual prompts to an LLM API. Which response is MOST appropriate?
- A. Disable logging on the API to reduce alert noise
- B. Ignore the alert because the spike is probably a legitimate marketing event
- C. Investigate for abuse and apply rate limiting and input controls
- D. Increase the token limit so legitimate requests are not rejected
Answer: C
Unusual prompt spikes may signal probing or abuse, so investigation with rate limits and input controls is appropriate. Ignoring alerts, disabling logs or raising limits worsens exposure.
Domain 1: AI Governance and Program Management
Q499 A marketing team publishes AI-generated images that closely resemble a photographer's copyrighted portfolio. Which risk should the organization be MOST concerned about?
- A. Excess storage costs for the large image library the team maintains, and the effort needed to archive the older assets
- B. Intellectual property infringement and uncertainty over who owns the generated output
- C. Lower search engine ranking of the corporate website
- D. Slow rendering of images on customer mobile devices
Answer: B
GenAI raises elevated copyright, ownership and IP concerns when outputs resemble protected works. Storage, rendering speed and search ranking are operational or marketing matters that carry no comparable legal exposure.
Domain 1: AI Governance and Program Management
Q500 Developers want to train a model on scraped web content. What should the organization do FIRST to manage IP risk?
- A. Train on the content and remove material only if the owners complain
- B. Confirm that the content may lawfully be used for training under licence or permitted terms
- C. Anonymize the file names so the sources cannot be identified later
- D. Rely on the model vendor to assume all copyright liability automatically, since its terms of service cover customer use
Answer: B
IP risk is managed by confirming rights to use the content before training. Training first and reacting to complaints, assuming vendor liability, or hiding sources do not resolve the underlying exposure.