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
Q301 A security team deploys prompt templates to block prompt injection. Which approach does this BEST represent?
- A. Securing AI systems
- B. Business continuity planning
- C. AI-enabled network analysis
- D. Procurement
Answer: A
Protecting a model from prompt injection is an example of securing AI itself. Using an algorithm to analyze network traffic would be AI for security. The other items are unrelated.
Domain 1: AI Governance and Program Management
Q302 Why is it MOST important that the AI security strategy extend beyond the IT department?
- A. Only IT staff use AI tools, so extending the strategy would duplicate effort elsewhere
- B. AI use occurs across the enterprise, including outside IT, and risk spans the whole life cycle
- C. Other departments cannot be secured because they lack technical staff to implement controls
- D. Regulators assign AI security responsibility to the business units rather than to IT
Answer: B
A robust program should protect all AI solutions across the enterprise and the AI life cycle, not just those used by IT. Treating IT as the only user ignores shadow AI and business-unit tools.
Domain 1: AI Governance and Program Management
Q303 Which action is MOST likely to gain support for an AI security program from stakeholders affected by AI?
- A. Defer stakeholder engagement until after the controls are fully implemented
- B. Limit communication to the board so that staff concerns do not delay delivery
- C. Present the program as a cost-cutting initiative without discussing role changes
- D. Address workplace changes such as job displacement, reskilling and revised roles
Answer: D
To gain support, the program should consider changes in the workplace such as job displacement, reskilling, upskilling and revised roles. Ignoring concerns reduces buy-in, and narrow communication does not build support.
Domain 1: AI Governance and Program Management
Q304 Which metric type would the program MOST directly use to show alignment with organizational objectives?
- A. Key performance indicators linked to business outcomes
- B. Employee parking utilization and visitor badge issuance across all offices
- C. Office temperature and energy readings collected by building systems
- D. Counts of printer errors and help desk tickets for office equipment
Answer: A
Program metrics should connect to organizational objectives through KPIs and measures that stakeholders understand. Parking and printer data do not demonstrate alignment.
Domain 1: AI Governance and Program Management
Q305 A network team wants AI to help detect threats earlier. Which use is MOST consistent with AI-enabled network security?
- A. Analyzing traffic patterns to detect anomalies
- B. Disabling log collection to reduce storage and processing costs
- C. Replacing perimeter firewalls with conversational chatbots
- D. Allowing all inbound connections to avoid false positive alerts
Answer: A
AI can analyze network traffic patterns to enhance detection and prevention of security events. Removing firewalls, logs or restrictions reduces security. AI augments, rather than replaces, core controls.
Domain 1: AI Governance and Program Management
Q306 An AI tool helps a risk manager evaluate vendors across large volumes of data. Which area does this BEST represent?
- A. Payroll processing and benefit calculations for employees
- B. Physical access control for data center facilities and offices
- C. Hardware maintenance scheduling for servers and end-user devices
- D. AI-enabled supply chain and vendor management
Answer: D
AI can support supply chain and vendor management by analyzing large amounts of vendor data. Physical security, hardware and payroll are not described this way.
Domain 1: AI Governance and Program Management
Q307 An incident response program was designed for conventional IT incidents. Which AI-specific scenario is MOST likely to expose its gaps?
- A. A printer outage that stops document production in one office
- B. A power loss that interrupts operations in a single building
- C. Disclosure of personal data from a training dataset
- D. A forgotten password that locks an employee out of a workstation
Answer: C
Programs have often not been adapted for AI incidents such as disclosure of confidential or personal training data, harmful output abuse, and hallucinated or biased predictions. Passwords and printer outages are conventional events.
Domain 1: AI Governance and Program Management
Q308 Where should debate over using AI to respond to an incident PRIMARILY take place?
- A. After the incident only, once all facts are confirmed by the vendor
- B. During the incident, under time pressure, so decisions reflect live conditions
- C. Outside of active incident response, in advance
- D. Only with the AI vendor, which should make the final decision
Answer: C
The manual says debate and decision-making over the use of AI should occur outside active incident response efforts. Deciding in the middle of an incident wastes time. Advance policy enables fast action.
Domain 1: AI Governance and Program Management
Q309 Which additional member is MOST useful on an AI incident response team to interpret anomalous model behavior?
- A. Event planners coordinating corporate meetings and travel
- B. Facilities managers responsible for building operations
- C. Travel coordinators who arrange staff trips
- D. Data engineers and scientists
Answer: D
Data engineers and scientists took part in preprocessing and training and can help interpret anomalous output and behavior. Privacy experts and ethicists also help. Facilities or event staff add little.
Domain 1: AI Governance and Program Management
Q310 How should an AI tabletop exercise PRIMARILY differ from a conventional breach exercise?
- A. Poisoning is analyzed in exactly the same way as a conventional network breach
- B. Data engineering and science teams must be excluded to preserve independence
- C. Evaluating data poisoning needs dataset access and collaboration with data teams
- D. AI incidents never require forensic tools or log analysis during investigation
Answer: C
Procedures for a system breach differ from those for data poisoning, which require access to or a copy of the data lake and collaboration with data engineering and science teams. AI-specific exercises build experience with dataset exploration.