AAIA exam content outline: domains, topics, tasks

The AAIA exam content outline is ISACA's map of the exam: three domains, each with a share of the 90 questions, 17 topics within them, and 23 supporting tasks that describe what a certified AI auditor does. It is short, two pages in the candidate guide, and it is the one document that tells you what the exam covers. This guide goes through it domain by domain, explains what the shares mean and what they do not, groups the tasks by what they ask of you, and points out the two small differences between the guide's version and ISACA's web page.

The short answer

• Three domains: AI Governance and Risk (33%), AI Operations (46%) and AI Auditing Tools and Techniques (21%).
• 17 topics: five in the first domain, seven in the second and five in the third, coded 1A to 3E.
• 23 supporting tasks, of which 21 begin with "Evaluate".
• Only the domains carry weights. ISACA gives no share to any topic or task.
• The shares describe the exam, not the score. "Domain percentages indicate the portion of the exam that reflects that domain content and are not used to calculate exam scores."
• The current outline is dated June 2025, as the guide prints it.

Where the outline lives

ISACA publishes the outline in two places. The candidate guide prints it as an appendix headed "AAIA Advanced in AI Audit Examination Content Outline (June 2025)", with the domains, their shares, the 17 topics and the 23 tasks. ISACA's AAIA exam content outline page gives the same domains and topics with a short description of each domain, and introduces the exam this way: it "consists of 90 questions covering three areas, all testing your ability to address real-world AI related opportunities and challenges."

ISACA says where the content comes from: "The domains, subtopics and tasks are the results of extensive research, feedback and validation from subject matter experts and prominent industry leaders from around the globe."

The two versions agree on substance. They differ in two small ways: the guide heads the task list "Supporting Tasks" and numbers it, while the web page lists the same 23 tasks, in a different order, under "OTHER SKILLS TESTED"; and the web page sets some topic names in capitals. Neither difference changes what is examined.

Domain 1: AI Governance and Risk (33%)

ISACA's description: "This domain demonstrates your ability to advise stakeholders on implementing AI solutions to meet organizational strategic goals, creating ethical and responsible AI policy and governance practices, and mitigating implementation risks, including leading practices in data governance, privacy and security."

Its five topics:

• 1A, AI Models, Considerations, and Requirements. What AI systems are, how they are built and used, and what an organization must decide before adopting one.
• 1B, AI Governance and Program Management. Who directs, owns and oversees AI in an organization, and how a program turns policy into practice.
• 1C, AI Risk Management. Identifying, assessing and treating the risks AI brings, from bias to dependence on a vendor.
• 1D, Privacy and Data Governance Programs. How privacy law and data governance apply when data feeds models.
• 1E, Leading Practices, Ethics, Regulations, and Standards for AI. The laws, standards and principles an auditor uses as criteria.

At 33%, this domain is about 30 of the 90 questions. It reads as the auditor's view of the organization around the AI: is there governance, does it work, are the risks and the obligations understood.

Domain 2: AI Operations (46%)

ISACA's description: "This domain confirms your skill in assessing an organization's risk profile and the rewards and consequences of AI implementations while ensuring the organization is operationally ready for successful adoption of this powerful technology."

Its seven topics:

• 2A, Data Management specific to AI. Sourcing, quality, labels, lineage and the data a model learns from.
• 2B, AI Solution Development Methodologies and Lifecycle. How models are built, validated, approved and documented.
• 2C, Change Management specific to AI. Retraining, new versions and vendor updates as changes to control.
• 2D, Supervision of AI Solutions (e.g., outputs, impacts, and decisions). Monitoring, human oversight and decision records once a model is live.
• 2E, Testing Techniques for AI Solutions. How AI systems are tested, and what an auditor can test or repeat.
• 2F, Threats and Vulnerabilities specific to AI. Attacks on models and their data, from poisoning to prompt injection.
• 2G, Incident Response Management specific to AI. Detecting, containing and learning from AI failures.

At 46%, nearly half the exam, about 41 questions, this is the largest domain by a distance. It follows a model through its life, from the data it learns from to the incident that takes it offline, and asks at each stage whether the organization has the controls that stage needs.

Domain 3: AI Auditing Tools and Techniques (21%)

ISACA's description: "This domain focuses on optimizing audit outcomes through innovation while highlighting your knowledge of audit techniques tailored to AI systems and your use of AI-enabled tools to streamline audit efficiency and provide faster, quality insight."

Its five topics:

• 3A, Audit Planning and Design. Scoping an AI audit by risk, setting criteria and planning the work.
• 3B, Audit Testing and Sampling Methodologies. Choosing tests, samples and populations for AI systems.
• 3C, Audit Evidence Collection Techniques. What counts as reliable evidence about a model, and how to keep it.
• 3D, Audit Data Quality and Data Analytics. Using analytics on AI data, and trusting your own analysis.
• 3E, AI Audit Outputs and Reports. Writing findings, recommendations and reports about AI.

At 21%, about 19 questions, this is the smallest domain, and the most familiar to an experienced auditor: planning, testing, evidence and reporting, applied to systems that learn. It also covers using AI in your own audit work.

What the shares mean, and what they do not

The three percentages tell you how the exam is weighted, not how it is scored. The guide is explicit: "Domain percentages indicate the portion of the exam that reflects that domain content and are not used to calculate exam scores." Your score is "based on the total number of exam items answered correctly, regardless of domain", and domain results on your report "are provided for informational purposes only." AAIA exam format and scoring covers the rest of the scoring.

Two practical points follow. First, time spent should roughly follow the shares: an hour on AI Operations is worth more than an hour on any other domain. Second, ISACA does not weight the topics, so do not read the outline as a promise that each topic gets the same number of questions. Within a domain, ISACA says nothing about how its questions spread across the topics. To plan study time by topic, our course splits each domain's share across its topics as evenly as whole numbers allow, and says so on every page where that split appears.

The 23 supporting tasks

The tasks describe what an AAIA holder does. Twenty-one of the 23 begin with "Evaluate"; the other two begin "Utilize" and "Analyze". Grouped by what they ask of you, in our words, with a phrase from each in ISACA's:

• AI inside the audit itself. Using AI to improve auditing, and judging the risk of doing so: "Utilize AI solutions to enhance audit processes, including planning, execution, and reporting."
• Advising on AI's impact. On the organization, on people and on the workforce, including the role of "AI decision-making systems on the organization and stakeholders" and how to address "AI-related workforce impacts, training, and education".
• Governance in practice. Policies and their legal compliance, metrics such as KPIs and KRIs, "defined ownership of AI-related risk, controls, procedures, decisions, and standards", and awareness programs.
• Data and privacy programs. The organization's data governance and privacy programs specific to AI, and "data input requirements for AI models (e.g., data appropriateness, bias, privacy)".
• Operational programs. Problem and incident management, change management, configuration management, threat and vulnerability management, and identity and access management, each "specific to AI".
• Vendors. "Evaluate vendors and supply chain management programs specific to AI solutions."
• The system itself. Controls, assessed for "the design and effectiveness of controls specific to AI", business requirements and enterprise architecture, "the AI solution lifecycle (e.g., design, development, deployment, monitoring, and decommissioning)", and "algorithms and models".

ISACA's full numbered list is on page 23 of the candidate guide and on its outline page. The guide lists the tasks after the domains and topics without assigning any task to a domain or topic.

What the outline does not name

The outline names topics, not documents. Topic 1E covers "Leading Practices, Ethics, Regulations, and Standards for AI", but the outline does not list which laws, standards or frameworks, and no domain names a specific one. That leaves the question of what to study for 1E to the candidate.

Our lessons answer it with the instruments an AI auditor meets most. Counting our 18 lessons, the EU AI Act appears in 12, the NIST AI Risk Management Framework in 7, ISO/IEC 42001 in 6, the GDPR in 4, the OWASP Top 10 for LLM Applications in 3 and MITRE ATLAS in 2. Those choices are ours, not ISACA's, and where a law has changed since a lesson was written, the lesson gives the rule now in force.

Using the outline to plan your study

The outline is short, but it can carry a whole study plan if you use it in four ways:

1. Weight your time by the domains. Roughly a third of your hours on AI Governance and Risk, nearly half on AI Operations and a fifth on AI Auditing Tools and Techniques. Familiar audit topics in Domain 3 can tempt a quick skim, and governance in Domain 1 is easy to read about at length, but the shares say Domain 2 deserves the most.
2. Cover every topic. With no weights below the domains, there is no safe topic to skip. A topic you leave out could carry any number of the questions in its domain.
3. Turn each task into a self-check. For a task such as evaluating "the organization's change management program specific to AI", ask yourself three things: what evidence you would request, what a working control looks like, and what finding you would write if it were missing. If you cannot answer all three, that topic needs more work.
4. Practice by domain, then mixed. Answer questions topic by topic while you learn, then in mixed sets, because the exam does not tell you which domain a question belongs to.

How to study for the AAIA turns these four into a four-week plan.

How the outline shows up in a question

Because the tasks are written from the auditor's chair, many questions are about roles and structures as much as technology. AAIA exam questions has three more, with full rationales. A Domain 1 question might look like this:

Where to start

The outline is the best study plan you will get from ISACA: weight your time by the domains, cover every topic, and practice the auditor's judgment the tasks describe. Our course follows it exactly, with a lesson for each of the 17 topics (two for 1A) and questions on every one. A free account opens AI models and requirements, topic 1A, in full; the practice page shows how our 619 questions spread across the topics; and full access is a single payment with nothing to renew. If you are not sure you qualify, see AAIA certification requirements.