Advanced in AI Security Management

A credential built on top of a credential

You can't walk in cold. ISACA requires an active CISM or CISSP before you're even eligible to register for the AAISM exam, which makes this one of the few certifications where every single test-taker is already a working security manager, not someone breaking into the field. That changes what "practice questions" need to do here — you don't need refreshers on what a risk register is; you need scenario reps on how AI changes the answer to questions you already know how to ask.

What the AAISM exam actually tests

The official AAISM exam is 90 multiple-choice questions across three domains, and ISACA has published the exact weighting on its official content outline — this isn't a case of guessing at proportions.

Domain 1 — AI Governance and Program Management (31%) This is the policy and program side: standing up AI-specific security policies, managing the AI asset and data lifecycle, running incident response for AI-specific events, and advising stakeholders through the actual regulatory frameworks that now touch AI deployments. Expect questions built around the "provider vs. deployer" distinction — a recurring theme in ISACA's own materials because the security obligations genuinely differ depending on which side of that line your organization sits on.

Domain 2 — AI Risk Management (31%) Threat and vulnerability management for AI systems specifically, AI risk assessment and treatment, and vendor/supply chain risk when a third party's model or dataset is now embedded in your environment. The "buy vs. build" decision shows up repeatedly here, because the risk calculus for a licensed foundation model looks nothing like the risk calculus for something trained in-house.

Domain 3 — AI Technologies and Controls (38%) The largest domain by a clear margin, and the one people most consistently underestimate. It covers AI security architecture, the AI lifecycle (model selection, training, validation), data management controls, and the privacy/ethics/trust and safety controls layer that doesn't have a clean analog in traditional CISM-style material. This domain is where candidates with strong governance backgrounds and weaker hands-on AI exposure tend to lose the most points.

If you want the granular breakdown, ISACA also publishes 22 supporting tasks under the content outline — things like "conduct AI impact assessments," "design and implement testing and vulnerability management of AI solutions," and "advise on security risk related to the AI solution development lifecycle." Treat that task list as your real syllabus; it's more specific than any third-party study guide's chapter headings.

Who's actually sitting for AAISM exam

Because of the CISM/CISSP prerequisite, the AAISM candidate pool skews toward people already carrying titles like security manager, CISO, cybersecurity architect, or compliance officer — professionals who've been told, explicitly or implicitly, "you now own AI risk for this organization" and need a credential that proves it. A smaller group comes from GRC and AI governance roles who hold CISSP as a secondary credential and are angling to lead AI oversight programs rather than build AI systems themselves. It's worth being clear about what this exam is not: it's not a data science or ML engineering credential, and questions won't ask you to reason about model architectures the way a technical AI certification would. It tests management and oversight judgment applied to AI-specific problems.

AAISM Exam format details that actually matter for practice

You get 2.5 hours for 90 questions — about 100 seconds per question on average, which is more breathing room than CISM's pacing but still tight once you hit the longer scenario-based items in Domain 3. The exam uses ISACA's 200–800 scaled scoring system, and you need 450 to pass. It's administered at PSI test centers or via remote proctoring — except for candidates in India, Mainland China, and Hong Kong, who must test at a physical center; there's no remote option available in those regions.

One practical detail that trips people up during scheduling rather than during the exam itself: once you register, you have a six-month eligibility window to actually sit for the exam. That's shorter than some candidates expect, and it means your practice-test timeline should be locked in before you pay the registration fee, not after.

How to use this practice test effectively

Don't run through these questions as a single pass-fail simulation and call it done. Because Domain 3 carries 38% of the exam and covers the least overlap with general CISM/CISSP knowledge, weight your practice accordingly — if you're only doing proportional review by domain count (three domains, roughly equal attention), you're under-preparing for more than a third of the actual exam. Use the practice questions to specifically test whether you can articulate the difference between provider and deployer obligations, and whether you can reason through a buy-vs-build scenario the way ISACA frames it — as a governance decision with security consequences, not a procurement question.

Retake the practice set after a week's gap and watch which domain your score drops in. That's usually a more honest signal than your first-pass score, because AI governance vocabulary is easy to recognize but harder to apply cold under time pressure.

Common mistakes candidates make on AAISM exam

The most frequent misstep is treating AAISM like an extension of CISM with an "AI" label slapped on the same underlying logic. The exam specifically probes AI-native distinctions — data lifecycle management for training data versus production data, human oversight requirements for AI outputs, explainability and robustness as security concerns rather than just technical footnotes — that don't have a direct CISM equivalent to fall back on. Candidates who skim the AI-specific vocabulary and assume their existing risk-management instincts will carry them through Domain 3 tend to be the ones surprised by their score.

A second common mistake is under-preparing for the incident response material in Domain 1. AI incident handling — containment, notification, escalation, eradication, recovery — sounds like standard security incident response until you factor in the additional regulatory reporting obligations that apply specifically to AI-related incidents in certain jurisdictions. Questions here often hinge on that added layer, not the base incident response process itself.

Study tips specific to AAISM exam

Read ISACA's own AAISM Exam Content Outline before you touch a third-party guide. The domain weightings (31/31/38) and the 22 supporting tasks are published directly by ISACA, and mapping your study time against that list — rather than a generic "AI security" curriculum — keeps you focused on what's actually tested rather than what's generally interesting about AI risk.

Because this exam sits on top of CISM or CISSP knowledge you already hold, resist the urge to re-study foundational security management concepts. Spend that time instead on the handful of concepts that are genuinely new: provider vs. deployer obligations, AI asset/data lifecycle distinctions, AI-specific vendor and supply chain risk, and the privacy/ethics/trust and safety control category that Domain 3 introduces. These are the concepts practice questions should be stress-testing, not restating basic risk management principles you'd already pass a CISM exam on.

Where to go from here

Work through the free questions below to get a feel for how AAISM frames governance and risk scenarios around AI specifically, then check which domain gave you the most trouble. If it's Domain 3, that's normal — it's the largest domain and the one furthest from traditional security management material — and it's where focused review time will move your score the most before exam day.