English
中文English

Certified in Risk and Information Systems Control Practice Exam — AAISM：Advanced in Al Security Management

1. The question bank is cloud‑connected and updates automatically; no manual re‑acquisition is required.

2. Start practicing right after activating the question bank. It supports simultaneous use on websites and mini‑programs, with one‑click bilingual switching for each question.

3. Functions include online practice, mock tests, note‑taking, wrong‑question recording, etc., valid for one year.

4. Recommended practice order: Turn on review mode to browse questions → Complete sequential practice → Take mock exams for pre‑test self‑assessment.

5. Activation codes can be purchased by clicking Buy Now on the right or via our official Tmall flagship store.

6. For inquiries, contact customer service through mini‑program, WeChat, WhatsApp or LINE.

Exam information

I. Basic Exam Information

- Exam Name: ISACA Advanced in AI Security Management™ (AAISM™)

- Exam Languages: Only English is available globally at present. No Chinese version has been announced. Candidates in Mainland China may only take the exam in English for the time being.

- Registration Eligibility: There are no restrictions for exam registration, and all individuals may sign up for the exam. A formal AAISM credential application is required upon passing the exam. Credential prerequisites:

 • Hold an active CISM or CISSP credential (either one is acceptable)

 • Possess practical work experience in security or consulting fields

 • Have relevant experience in evaluating, implementing and maintaining security controls

 • Agree to and abide by the ISACA Code of Professional Ethics

- Exam Fees: USD 575 for ISACA members; USD 760 for non-members. A USD 50 credential application fee is required after passing the exam.

- Exam Duration: 2.5 hours (150 minutes), covering all 90 exam questions.

- Exam Format: ISACA offers two standard exam modes worldwide: remote-proctored online exam (video proctoring via PSI) and in-person computer-based testing (CBT).

 Available exam modes by region:

 • Hong Kong: In-person CBT only

 • Taiwan: In-person CBT and online exam

 • Macau: In-person CBT only

 • Mainland China: In-person CBT arranged by ISACA-authorized partners

 • Overseas regions: Free to choose between remote proctoring and in-person CBT

- Exam Questions: 90 scenario-based multiple-choice questions. All questions are scored with no unscored pre-test items.

- Scoring Details:

 • A preliminary pass/fail result is provided immediately after the exam.

 • Official exam results will be sent via email within 10 working days.

 • Passing score: 450 points (total score: 800 points; scaled scoring range: 200 – 800).

- Exam Eligibility Validity: After successful registration, candidates must schedule and complete the exam within 6 months. Eligibility will expire thereafter, and re-registration and payment will be required.


II. Detailed Exam Content (Latest Exam Outline)

The AAISM exam consists of three core knowledge domains with a total weight of 100%. It assesses the professional competencies and practical capabilities of AI security management practitioners.


1. AI Governance and Project Management (31%): AI strategy and business alignment, AI governance frameworks, policy and standard formulation, AI project management, stakeholder management, regulatory compliance requirements (e.g., EU AI Act), assessment of ethical and social impacts of AI.

2. AI Risk Management (34%): AI threat landscape analysis, risk identification and assessment, AI supply chain risk management, risk response strategies (avoidance, mitigation, transfer, acceptance), risk monitoring and reporting, AI-specific risks including data poisoning, model bias, prompt injection and more.

3. AI Technologies and Controls (35%): AI security architecture design, end-to-end data lifecycle security management, model security (training, testing, deployment and decommissioning), access control and identity management, security testing and validation for AI systems, incident response and recovery, evaluation and optimization of AI security controls.


Core Assessment Focus

Enterprise-level AI security governance capabilities, identification and mitigation of AI-specific risks, implementation of AI security technologies and controls, cross-departmental collaboration and stakeholder management capabilities.


III. Registration Process

1. Registration Process for Non-Mainland China Candidates

1. Visit the official ISACA website: https://www.isaca.org, create and log in to your MyISACA account.

2. Select the AAISM exam for registration, and fill in personal information as well as proof of relevant credentials.

3. Complete exam fee payment (USD 575 for members / USD 760 for non-members).

4. Upon successful payment, your 6-month exam eligibility period will take effect.

5. Schedule your exam time and location via PSI (remote proctoring or in-person CBT is optional).

6. Prepare valid identification documents and take the exam at the scheduled time.


2. Registration Process for Mainland China Candidates

1. Register via the official ISACA China website: https://www.isaca.net.cn or ISACA-authorized partners such as ZhongShen Audit Online and Saihu Academy.

2. Submit personal information and valid credential documentation, and complete registration with assistance from authorized institutions.

3. Pay the exam fee (USD 575 for members / USD 760 for non-members). All registration procedures will be processed uniformly by the institution.

4. Upon successful registration, your 6-month exam eligibility period will take effect.

5. Follow the links or guidelines provided by the institution to schedule your exam time and test center on the PSI platform. In-person CBT is the primary option in Mainland China.

6. Present valid identification documents (ID card or passport) on exam day.


IV. Supplementary Notes

1. Credential Validity: The AAISM credential is valid for 3 years. To maintain active status, holders must earn 120 Continuing Professional Education (CPE) credits within the validity period and pay the annual maintenance fee.

2. Retake Policy: Candidates who fail the exam must wait 30 days before retaking it. A 90-day waiting period is required after two consecutive failures. The retake fee is the same as the initial exam fee.

3. Exam Updates: Launched on August 19, 2025, this exam focuses on AI security governance, AI-specific risk management and implementation of AI security controls, which aligns with the core demands of enterprise AI security management.

4. Distinction from Other AI-Focused Credentials: AAISM targets AI security management and is designed for security leaders. AAIA (Advanced in AI Audit) focuses on AI audit for audit professionals. AAIR (Advanced in AI Risk) concentrates on AI risk management for risk practitioners.


Wish all candidates every success in the exam!

Sample questions

Certified in Risk and Information Systems Control · Q1
Question #1 An AI research team is developing a natural language processing model that relies on several open-source libraries. Which of the following is the team's BEST course of action to ensure the integrity of the software packages used?
  • A.
    Maintain a list of frequently used libraries to ensure consistent application in projects.
  • B.
    Retrain the model regularly to handle package and library updates.
  • C.
    Scan the packages and libraries for malware prior to installation.
  • D.
    Use the latest version of all libraries from public repositories.

Answer: C

The correct answer is C. Ensuring the integrity of open source software packages is a core requirement of the All AAISM (AI Assurance, Security, and Integrity Management) certification's software supply chain security domain. Open source libraries are a common vector for supply chain attacks, including malware injection, typosquatting, and repository tampering, which can compromise the confidentiality, integrity, and availability of the associated NLP model. Scanning packages for malware and tamper indicators prior to installation directly validates that the libraries are unmodified, free of malicious content, and match their expected legitimate state, addressing the core integrity requirement stated in the scenario. This control prevents compromised packages from being integrated into the development or production environment, reducing risk of model tampering, data exfiltration, or incorrect model outputs caused by malicious library code. Option Analysis: A. Incorrect. Maintaining a list of frequently used libraries supports consistency across projects but does not address integrity. Even widely used, trusted libraries can have compromised versions available in public repositories, so a curated list alone does not verify that the specific package instance being installed is unmodified or free of malware, per AAISM security requirements. B. Incorrect. Regular model retraining addresses model performance drift and adaptation to new data distributions, but it has no connection to the integrity of underlying software libraries. Retraining does not detect or mitigate tampered or malicious packages, so it is irrelevant to the stated requirement of ensuring software package integrity. C. Correct. Pre-installation scanning of open source packages for malware, known vulnerabilities, and tamper evidence is a mandated AAISM control for AI software supply chain integrity. This step directly verifies that the library packages being used have not been altered by malicious actors and do not contain hidden harmful code, fulfilling the core integrity requirement for third-party dependencies in AI system development. D. Incorrect. Using the latest version of libraries from public repositories does not guarantee integrity. Latest releases can still contain undisclosed vulnerabilities, intentionally inserted malware, or be subject to repository compromise. Pulling packages directly from unvetted public repositories also introduces risk of man-in-the-middle attacks, so this practice does not meet AAISM integrity validation requirements. Key Concepts: 1. AI Software Supply Chain Security: This core AAISM knowledge domain covers end-to-end security controls for all components used in AI system development, including open source libraries, to prevent unauthorized modification, malware injection, and supply chain compromise that can degrade AI system integrity. 2. Open Source Software (OSS) Integrity Validation: An AAISM-mandated control that requires verification of the authenticity and security of all OSS dependencies prior to integration, to mitigate risks from compromised public repositories, typosquatting attacks, and tampered package releases. 3. Pre-Deployment Dependency Scanning: A core AAISM security practice that requires scanning all third-party software dependencies, including libraries, for vulnerabilities, malware, and configuration issues before they are deployed in development or production AI environments. References: NIST AI Risk Management Framework, https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf OWASP Top 10 for Large Language Model Applications, https://owasp.org/www-project-top-10-for-large-language-model-applications/
Certified in Risk and Information Systems Control · Q2
Question #2 An organization plans to apply an AI system to its business, but developers find it difficult to predict system results due to lack of visibility to the inner workings of the AI model. Which of the following is the GREATEST challenge associated with this situation?
  • A.
    Assigning a risk owner who is responsible for system uptime and performance
  • B.
    Continuing operations to meet expected AI security requirements
  • C.
    Determining average turnaround time for AI transaction completion
  • D.
    Gaining the trust of end users through explainability and transparency

Answer: D

The scenario describes the core AI "black box" problem, where inner model workings are opaque to developers and stakeholders. Per All AAISM Questions certification domain knowledge, the greatest impact of opaque AI models is the inability to demonstrate valid, consistent output generation to end users. Trust is a foundational requirement for successful AI deployment, as without user confidence in system outputs, the AI system will not be adopted or used appropriately, negating all business value from the implementation. The lack of visibility directly prevents the organization from providing the explainability and transparency required to build this trust, making this the highest priority challenge in the scenario. Option Analysis: A. Incorrect. Assigning a risk owner for system uptime and performance is a standard operational risk management task that is not dependent on visibility into AI model inner workings. Risk owners for operational performance can be designated for any system, including black box AI models, as their responsibility focuses on availability and operational throughput rather than model decision logic, so this is not a challenge tied to the scenario's core issue. B. Incorrect. AI security requirements including data protection, access control, and adversarial resilience can be implemented for opaque AI models without visibility into inner model logic. Per AAISM frameworks, core AI security controls apply to the supporting infrastructure, training data, and access layers, which are independent of model interpretability, so this is not the greatest challenge in this scenario. C. Incorrect. Average transaction turnaround time is an operational performance metric that is measured via standard monitoring tools, regardless of whether the AI model's inner workings are visible. Turnaround time can be tracked, benchmarked, and optimized even for fully opaque black box models, so this is unrelated to the core challenge described. D. Correct. Per AAISM certification knowledge, explainability and transparency are mandatory control objectives for AI systems to build end user trust. When model inner workings are not visible, the organization cannot provide clear, understandable explanations of how system outputs are generated, or validate that outputs are free from bias or error for end users. This directly erodes end user trust, which is the greatest barrier to successful AI deployment, as low trust leads to low adoption, misuse, or full rejection of the system by end users. Key Concepts: 1. AI Black Box Problem: This core AAISM concept refers to opaque AI models such as deep learning networks or complex ensemble models where the logic used to generate outputs cannot be easily interpreted by human stakeholders. This opacity prevents auditing, explanation, and validation of model decisions, creating significant trust and compliance risks. 2. AI Explainability and Transparency: A foundational AAISM control domain that requires AI system outputs and decision processes to be explainable in plain language to all relevant stakeholders including end users, regulators, and business leaders. Transparency includes disclosure of model training data, intended use cases, and performance limitations, which is not possible without visibility into inner model workings. 3. AI Stakeholder Trust: AAISM identifies end user and stakeholder trust as a primary key performance indicator for successful AI implementation. Even technically high-performing AI systems fail to deliver business value if users do not trust outputs enough to act on recommendations or use the system as intended. References: NIST AI Risk Management Framework, https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf OECD Principles on Artificial Intelligence, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0449
Certified in Risk and Information Systems Control · Q3
Question #3 Which of the following is MOST important to consider when validating a third-party AI tool?
  • A.
    Terms and conditions
  • B.
    Roundtable testing
  • C.
    Right to audit
  • D.
    Industry analysis and certifications

Answer: C

The correct answer is C. Right to audit. When validating a third-party AI tool, organizations remain fully accountable for all risks associated with the tool's use, including biases, data privacy violations, insecure model architecture, non-compliance with regulatory requirements, and inaccurate outputs, per AAISM third-party AI governance standards. The right to audit is the only mechanism that enables an organization to conduct independent, granular, use case-specific verification of the third-party AI system's end-to-end operations, rather than relying solely on vendor-provided disclosures, point-in-time certifications, or surface-level testing. This capability is foundational to validating that the tool aligns with the organization's risk tolerance, security policies, ethical AI principles, and mandatory regulatory obligations. Option Analysis: A. Terms and conditions: Incorrect. While clear terms and conditions define contractual obligations, usage rights, and liability boundaries between the organization and vendor, they are a legal document that does not enable direct validation of the AI tool's actual performance, security, or compliance. They are a necessary pre-contractual step but not the most important consideration for validation activities. B. Roundtable testing: Incorrect. Roundtable testing, or cross-functional user acceptance testing, only assesses the AI tool's surface-level outputs against basic functional requirements for the organization's use case. It cannot evaluate underlying risks such as biased training data, unpermitted data usage, hidden model vulnerabilities, or non-compliant data processing practices, all of which require access to the vendor's internal AI system documentation and processes that only audit rights can provide. C. Right to audit: Correct. This aligns with core AAISM AI risk management guidance which states that organizations cannot outsource accountability for third-party AI risks. The right to audit grants the organization permission to review the vendor's AI training data practices, model development lifecycle, security controls, incident response processes, and compliance records, enabling end-to-end validation of the tool's suitability and adherence to all internal and external requirements. D. Industry analysis and certifications: Incorrect. Third-party certifications and industry analyst assessments provide general, point-in-time assurance of a vendor's baseline maturity, but they are rarely tailored to the organization's specific use case, risk profile, or regulatory obligations. These resources are useful for initial vendor screening but are not sufficient for comprehensive, ongoing validation of the AI tool, as they do not grant access to current, granular system data required for full validation. Key Concepts: 1. Third-Party AI Accountability: A core AAISM principle stating that organizations that deploy or use third-party AI tools retain full accountability for all harms, compliance violations, and risks associated with those tools, requiring independent verification mechanisms rather than blind trust in vendors. 2. AI Validation Lifecycle: The AAISM-defined end-to-end process for confirming AI tools meet functional, security, ethical, and compliance requirements, which includes both pre-deployment assessment and ongoing monitoring, both of which are dependent on access to vendor data enabled by audit rights. 3. AI Regulatory Compliance Obligations: Global AI regulations including the EU AI Act and US state AI governance rules mandate that organizations using high-risk AI systems maintain full records of all system components, including third-party tools, which requires audit access to vendor systems to demonstrate compliance during regulatory audits. References: NIST AI Risk Management Framework, https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf European Commission Artificial Intelligence Act
Certified in Risk and Information Systems Control · Q4
Question #4 After implementing a third-party generative AI tool, an organization learns about new regulations related to how organizations use AI. Which of the following would be the BEST justification for the organization to decide not to comply?
  • A.
    The AI tool is widely used within the industry.
  • B.
    The AI tool is regularly audited.
  • C.
    The risk is within the organization's risk appetite.
  • D.
    The cost of noncompliance was not determined.

Answer: C

The scenario describes an organization that has deployed a third-party generative AI tool and is now evaluating compliance with newly enacted AI use regulations. Per core AAISM AI governance and risk management principles, the only valid, governance-aligned justification for a deliberate decision not to comply with a new regulation is a formal, documented risk assessment that confirms the residual risk of noncompliance (including potential fines, reputational damage, and operational harm) falls within the organization's pre-defined, board-approved risk appetite. Organizations retain full accountability for all risks associated with third-party AI tools, so decisions related to regulatory compliance must be rooted in structured risk evaluation rather than informal or unvalidated factors. Option Analysis: A. Incorrect. Widespread industry use of a third-party AI tool does not absolve an organization of its unique regulatory compliance obligations. Industry common practice is not a recognized legal or governance defense against noncompliance penalties, and does not account for the organization's specific operating sector, jurisdiction, or risk profile as required by AAISM standards. B. Incorrect. Regular audits of the AI tool only provide assurance of the tool's adherence to existing internal controls or past requirements, but do not address gaps related to newly enacted regulations. Audits identify risk gaps, they do not authorize noncompliance with updated regulatory mandates, so this cannot justify a decision to avoid compliance. C. Correct. Risk appetite is defined as the amount of risk an organization is willing to accept in pursuit of its business objectives, per AAISM AI governance frameworks. If the assessed risk of noncompliance with the new AI regulation is formally validated to be within the organization's approved risk appetite threshold, this is the only valid, risk-aligned justification for a decision not to comply. D. Incorrect. If the cost and impact of noncompliance are undetermined, the organization cannot complete the required risk assessment to support an informed noncompliance decision. Undetermined risk means the organization lacks the necessary visibility to justify avoiding compliance, so this factor prevents rather than supports a noncompliance decision. Key Concepts: 1. Risk Appetite Alignment: This core AAISM governance principle requires all risk-related decisions, including regulatory compliance choices for AI systems, to be evaluated against the organization's formally documented, stakeholder-approved risk appetite to ensure alignment with business objectives and acceptable risk tolerance levels. 2. Third-Party AI Accountability: AAISM mandates that organizations retain full, end-to-end accountability for all risks associated with third-party generative AI tools, including regulatory compliance risks, regardless of the tool's vendor, industry adoption rate, or existing audit status. 3. AI Regulatory Risk Assessment: AAISM requires organizations to conduct regular, structured assessments of AI-related regulatory risks, quantifying potential financial, reputational, and operational impacts of noncompliance to support evidence-based decision-making, rather than relying on anecdotal or informal factors. References: NIST AI Risk Management Framework (AI RMF 1.0), https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf ISO/IEC 42001:2023 Artificial Intelligence Management Systems — Requirements
Certified in Risk and Information Systems Control · Q5
Question #5 Which of the following is the MOST important consideration when deciding how to compose an AI red team?
  • A.
    Resource availability
  • B.
    Time-to-market constraints
  • C.
    Skills matrix
  • D.
    AI use cases

Answer: C

The correct answer is C. Skills matrix. Composing an effective AI red team relies first on ensuring the team has the specialized, cross-functional competencies required to identify and simulate realistic adversarial attacks against AI systems, which are distinct from traditional cybersecurity red team requirements. The skills matrix maps all necessary capabilities for AI red team operations, including adversarial machine learning expertise, AI lifecycle knowledge, traditional penetration testing skills, data security competence, and relevant domain expertise, to team members. This ensures no critical capability gaps exist that would prevent the team from accurately assessing AI system vulnerabilities, making it the most foundational consideration for team composition per AAISM AI security standards. Option Analysis: A. Resource availability: Incorrect. Resource availability including budget, tools, and infrastructure is a practical operational consideration for executing red team activities, but it is not the primary factor for composing the team itself. Even with unlimited resources, a team lacking the required specialized AI security skills cannot deliver meaningful red team outcomes, so this is a secondary consideration. B. Time-to-market constraints: Incorrect. Time-to-market constraints impact the scope, timeline, and prioritization of red team assessments, but not the core composition of the team. Rushing team composition to meet time-to-market targets by skipping skills validation results in ineffective testing that misses critical AI vulnerabilities, so this is an operational constraint rather than a core team composition consideration. C. Skills matrix: Correct. Per AAISM AI red teaming standards, AI systems have unique attack surfaces including data poisoning, model evasion, prompt injection, and training data exfiltration that require specialized skills not typically held by traditional red teams. The skills matrix ensures all required competencies to test these unique attack surfaces are present in the team, which is the non-negotiable foundation of an effective AI red team. Without the right mix of skills, no adjustment to resources or timelines can enable the team to deliver valid AI security testing results. D. AI use cases: Incorrect. AI use cases inform what specific skills are required for a given red team engagement, but they are an input to building the skills matrix, not the primary consideration for team composition. The actual composition of the team is directly dependent on matching required skills to team members via the skills matrix, so this is a secondary input rather than the core consideration. Key Concepts: 1. AI Red Team Competency Requirements: This core AAISM knowledge domain specifies that AI red teams require a unique blend of adversarial machine learning, traditional cybersecurity, AI system lifecycle, and domain-specific skills to address AI-specific attack surfaces that are not covered by standard red team testing. 2. AI Security Testing Team Composition Framework: This AAISM concept prioritizes competency mapping via a skills matrix as the first step in building AI red teams, ahead of logistical or operational considerations, to ensure full coverage of all AI vulnerability classes during testing. 3. Adversarial Machine Learning (AML) Testing Mandates: This AAISM knowledge point outlines that specialized AML skills are a non-negotiable requirement for AI red teams, as standard penetration testing methodologies do not address AI-specific attack vectors. References: NIST AI Risk Management Framework: Core, https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf MITRE ATLAS Red Team Guide

FAQ

How many practice questions are available for Certified in Risk and Information Systems Control?

This question bank includes 284 Certified in Risk and Information Systems Control practice questions covering single and multiple choice, each with answers and explanations.

Are Certified in Risk and Information Systems Control practice questions available in Chinese and English?

Yes, Certified in Risk and Information Systems Control practice questions are provided in both Chinese and English.

Can I try Certified in Risk and Information Systems Control practice questions for free?

Yes. Free sample questions are available on this page, and the full question bank is available after signing up on Zhangxuetu.

Try practice questions Get the question bank

Related exams