AAISM Exam Questions: Format, Scenarios and Structure 2026

What the AAISM Exam Actually Tests

The Advanced in AI Security Management (AAISM) certification, governed by ISACA and launched on August 19, 2025, sits at the intersection of two fields that most organizations are still scrambling to reconcile: artificial intelligence deployment and enterprise security management. It does not test AI theory in isolation, nor does it test generic security frameworks. It tests the specific, applied judgment required to govern, risk-assess, and technically control AI systems inside a real organizational context.

That framing matters enormously for how you prepare. AAISM questions are not asking you to recite definitions. They are presenting you with a situation - a CISO reviewing a new generative AI procurement, a risk officer evaluating model drift alerts, a security architect choosing between control architectures - and asking what the most appropriate, defensible course of action is. That is a fundamentally different cognitive challenge than a knowledge-recall exam.

Because the credential is a post-professional certification with a hard prerequisite of an active CISM or CISSP, ISACA's baseline assumption is that you already understand enterprise security principles. AAISM layers AI-specific complexity on top of that foundation. Candidates who approach it as "CISM with AI vocabulary" will underperform. Candidates who treat it as a genuinely new credential with its own conceptual demands tend to find the scenario logic more accessible once they internalize the three domains.

Question Format: Scenario-Based, Not Recall-Based

Every one of the 90 questions is multiple-choice and scenario-based. ISACA's published guidance for AAISM explicitly frames the exam around real-world AI security management scenarios, which means each question stem will typically include a contextual setup - an organization, a system, a decision point - before asking what should be done, assessed, or prioritized.

What Makes a Question "Scenario-Based"?

In practice, scenario-based questions on a credential like AAISM tend to follow a structure where:

• The stem describes an organizational context (type of AI system, regulatory environment, stakeholder dynamic).
• A triggering event or decision point is introduced (an audit finding, a model anomaly, a vendor proposal, a governance gap).
• The four answer options each represent plausible responses - often all technically defensible - but only one reflects the best practice given the specific context.

This structure rewards candidates who can distinguish between what is technically possible, what is procedurally correct, and what is strategically optimal. A candidate who knows that AI systems require adversarial robustness testing but cannot apply that knowledge to a procurement scenario will still struggle. The exam rewards applied judgment, not encyclopedic recall.

Time Per Question

With 90 questions and a 150-minute time limit, the average available time is exactly 100 seconds per question. That is sufficient for a well-prepared candidate, but scenario stems tend to be longer than simple knowledge-check questions. Pacing discipline - especially not over-analyzing options - is a practical exam-day skill worth practicing. Using timed practice tests at aaismexam.com before exam day is one of the most direct ways to calibrate your pacing under realistic conditions.

Domain Breakdown and Weight Distribution

These percentages come directly from ISACA's published AAISM exam content outline. The approximate question counts above assume all 90 questions contribute equally to your score, which is the standard structure for ISACA credentials. Note that ISACA may include unscored pilot items in any exam administration; those do not affect your result but are indistinguishable during the exam.

Domain 1: AI Governance and Program Management (31%)

Domain 1 questions frequently involve a senior security professional advising an organization on how to structure oversight of AI. Scenarios might involve a newly formed AI ethics committee, a gap between an existing information security policy and the requirements of a deployed large language model, or a CISO preparing to present AI risk posture to the board. The tested skill is governance design and program management judgment, not technical AI knowledge.

Candidates with a CISM background will recognize familiar governance themes here, but the AI-specific wrinkles - accountability for model behavior, third-party AI vendor oversight, and the unique transparency challenges of probabilistic systems - require additional preparation beyond what CISM covers.

Domain 2: AI Risk Management (31%)

Domain 2 is where candidates with traditional risk management backgrounds sometimes struggle most. AI introduces categories of risk that have no clean analogue in conventional IT risk frameworks. A model that gradually drifts from its training distribution, producing subtly incorrect outputs over time, is not a system outage - but it may represent a material security and business risk. AAISM questions in this domain will test whether you can recognize those novel risk categories and apply appropriate management responses.

Understanding how to read and interpret the AAISM Scaled Score Explained: 200 to 800 Grading Guide can help candidates understand where domain-level performance gaps show up in their score reports and how to prioritize remediation before a retake.

Domain 3: AI Technologies and Controls (38%)

Domain 3 scenarios tend to be the most technically detailed on the exam. A question might describe a machine learning pipeline and ask which control best mitigates a data poisoning risk during training, or present a deployed generative AI API and ask what monitoring approach would most effectively detect anomalous usage patterns. These questions require both conceptual understanding of AI architectures and practical knowledge of how security controls map to AI-specific components.

Exam Structure and Registration Mechanics

Delivery Format and Testing Options

AAISM is delivered through PSI, ISACA's testing partner, with two delivery options: authorized PSI testing centers globally, or remote proctoring. There is an important geographic exception - candidates in India, Mainland China, and Hong Kong are required to test at a physical PSI center. Remote proctoring is not available in those locations.

The exam is available in two languages: English and Spanish. This makes AAISM one of the more accessible enterprise AI certifications for Spanish-speaking professionals, which is consistent with ISACA's established multilingual exam strategy across credentials like CISM and CISA.

Registration Window and Fee Structure

Once registered, candidates have a 12-month eligibility window to sit for the exam. The exam fee is $459 for ISACA members and $599 for non-members. After passing, there is a one-time $50 application processing fee. Ongoing maintenance costs are $20 per year for ISACA members or $35 per year for non-members.

The cost difference between member and non-member pricing ($140 on the exam fee alone) frequently makes an ISACA membership financially rational for candidates who plan to pursue or maintain multiple ISACA credentials.

Prerequisite Requirements

AAISM has a hard prerequisite: you must hold an active CISM (Certified Information Security Manager) or CISSP (Certified Information Systems Security Professional) at the time of application, and that underlying credential must remain active for the entire lifecycle of your AAISM certification. If your CISM or CISSP lapses, your AAISM status is affected. This is not a one-time prerequisite check - it is an ongoing structural dependency.

Scoring, Eligibility, and the Scaled Score System

AAISM uses a scaled score system with a range of 200 to 800. The passing score is 450. Scaled scoring means that raw correct-answer counts are converted to a normalized scale before your result is determined - a methodology ISACA uses across its certification portfolio to ensure score comparability across different exam administrations.

For a detailed explanation of how this conversion works and what it means for your preparation strategy, the AAISM Scaled Score Explained: 200 to 800 Grading Guide covers the mechanics thoroughly.

Certification is valid for three years. Maintenance requires a minimum of 10 CPE hours per year in AI-specialized topics, totaling at least 30 CPE hours across the three-year cycle. CPE hours must be AI-relevant - general information security CPE that does not involve AI topics will not satisfy this requirement. Adherence to ISACA's Code of Professional Ethics is also a continuous obligation.

Who Pursues AAISM and Why

Because AAISM requires an active CISM or CISSP, the candidate pool is by definition experienced security professionals. The organizations most likely to prioritize this credential span several sectors:

• Financial services firms deploying AI for fraud detection, credit decisioning, or algorithmic trading - where regulatory scrutiny of AI model risk is increasing rapidly.
• Healthcare organizations using AI for diagnostics, triage, or administrative automation - where data privacy, model accountability, and adverse outcome liability intersect.
• Technology companies building or integrating AI products - particularly those with enterprise customers who impose security and governance requirements on vendors.
• Consulting and advisory firms - where an AAISM designation signals specialized AI security competency to clients evaluating engagements.
• Government and defense organizations - where AI procurement, deployment, and oversight increasingly require structured security governance.

The credential is relevant for CISOs establishing enterprise AI governance programs, senior security architects designing AI-integrated infrastructure, risk officers building AI-specific risk frameworks, and auditors assessing AI system controls. Practice tests at aaismexam.com are built around the exact scenario types these professionals will encounter on exam day.

A Domain-Weighted Preparation Approach

Because AAISM's domain weights are unequal and its content is genuinely specialized, a flat study schedule - spending equal time on all three domains - is suboptimal. A domain-weighted approach that reflects the exam's actual structure will yield better results.

Frequently Asked Questions