ISACA • AI-Fundamentals

Artificial Intelligence Fundamentals Certificate Practice Test

Validates foundational knowledge of artificial intelligence, covering AI concepts, principles, potential uses, essential algorithms and software for AI applications, and AI-associated risks and ethical requirements.

Exam Details

Questions

600

Duration

120 minutes

Passing Score

65%

Difficulty

Foundational

Last Updated

Feb 2026

Exam Domain Breakdown

AI Concepts50%

AI Implementations50%

Exam Overview

The ISACA Artificial Intelligence Fundamentals Certificate validates foundational knowledge of artificial intelligence, covering core AI concepts, principles, practical applications, essential algorithms, and the risks and ethical considerations that accompany AI adoption. The credential is designed to help professionals navigate the rapidly evolving AI landscape by building a solid understanding of technologies such as machine learning, neural networks, large language models, computer vision, robotic process automation (RPA), and generative AI. It bridges conceptual understanding with applied knowledge, ensuring candidates can identify AI use cases, understand how AI tools and algorithms function, and align AI practices with governance and regulatory frameworks.

As part of ISACA's expanding AI credentialing ecosystem, the AI Fundamentals Certificate serves as a foundational entry point into more advanced ISACA AI credentials, including the Advanced in AI Audit (AAIA) and the Advanced in AI Security Management (AAISM). The certificate is globally recognized and backed by ISACA's reputation as a trusted authority in IT governance, risk, and security — an organization with over 185,000 members across more than 190 countries.

Official exam page

Who Should Take This Exam

This certificate is well-suited for students, recent graduates, and early-career professionals who are new to AI and want to establish a verifiable baseline of AI knowledge. It is equally valuable for experienced IT professionals, auditors, risk managers, compliance officers, and business analysts who need to understand AI concepts and their organizational implications without necessarily working in a technical AI role.

Professionals seeking to transition into AI-adjacent roles — such as AI governance, IT audit with an AI focus, or risk and compliance in organizations adopting AI — will find this credential a practical starting point. Teams and organizations looking to upskill staff on AI fundamentals and demonstrate collective AI competency to stakeholders will also benefit from this certificate.

Prerequisites

There are no formal prerequisites for the ISACA AI Fundamentals Certificate. Registration is open on a continuous basis with no eligibility restrictions, and candidates can schedule their exam as early as 48 hours after payment of registration fees.

While no prior AI or IT experience is required, candidates will benefit from basic familiarity with IT concepts and business processes. ISACA recommends using its official study guide and the self-guided online course — which includes performance-based labs covering topics such as machine learning models, security implementations of AI, and robotic process automation — to build the foundational knowledge needed to pass the exam.

Exam Format

The exam is a computer-based, remotely proctored, multiple-choice assessment consisting of 60 questions, with a time limit of 120 minutes. It is delivered online through ISACA's remote proctoring platform and can be scheduled at any time, providing candidates with scheduling flexibility. No in-person testing center is required.

The passing score is 65% (39 out of 60 questions correct). The exam registration fee is US $120 for ISACA members and US $144 for non-members. Eligibility established at registration is valid for twelve months, and candidates may schedule their testing appointment up to 90 days in advance.

Skills Measured

1.Domain 1 — AI Concepts (50%): Covers the foundational characteristics and terminology of artificial intelligence, including machine learning paradigms (supervised, unsupervised, and reinforcement learning), neural networks, large language models (LLMs), generative AI, computer vision, robotic process automation (RPA), statistical modeling, and regression analysis. Candidates must demonstrate understanding of how these technologies function and their core distinctions.
2.Domain 2 — AI Implementations (50%): Covers the practical principles and applications of AI, including essential algorithms and software used to build and deploy AI systems, real-world AI use cases across business and IT contexts, and performance-based skills such as working with machine learning models and security implementations of AI. Candidates must demonstrate the ability to identify how AI can be applied and leverage fundamental tools to explore AI possibilities, as well as understand associated risks, ethical requirements, and governance considerations that guide responsible AI deployment.

Study Tips

Use the official ISACA AI Fundamentals Study Guide (2nd Edition) as your primary resource — it is authored by global industry experts, maps directly to exam domains, and includes example questions with annotated answers aligned to the 65% passing threshold.
Complete the ISACA self-guided online course, which earns 5 CPE credits and blends knowledge-based learning with hands-on performance labs covering machine learning models, AI security implementations, and robotic process automation — topics that appear in the AI Implementations domain.
Download and thoroughly review the official ISACA Exam Candidate Guide before registering. It details exam rules, proctoring requirements, scoring methodology, and the retake policy, helping you avoid procedural surprises on exam day.
Since the exam is split equally between AI Concepts (50%) and AI Implementations (50%), allocate study time evenly between understanding theoretical AI principles and practicing applied scenarios. Do not over-focus on one domain at the expense of the other.
Practice with the ISACA Lab Package (4.5 CPE credits), which provides a virtual environment to interact with AI tools hands-on. Practical exposure to ML models and RPA reinforces conceptual understanding and prepares you for performance-based application questions.
Review ISACA's published resources on AI ethics, AI governance frameworks, and AI risk management — topics that cut across both exam domains. ISACA's whitepaper library and AI resource hub (isaca.org/resources/artificial-intelligence) offer free materials directly aligned to the certificate's scope.
Take at least one full-length timed practice exam under realistic conditions (2 hours, 60 questions) before your scheduled test date to calibrate your pacing. Aim to answer each question in under 2 minutes to ensure you complete the exam comfortably within the time limit.

Career Benefits

The ISACA AI Fundamentals Certificate positions holders as credibly literate in AI at a time when organizations across every sector are integrating AI into operations, governance, and risk management. It provides a competitive edge for roles such as IT auditor, risk manager, compliance analyst, AI governance consultant, and business analyst — particularly as organizations seek professionals who can evaluate AI systems, identify associated risks, and ensure responsible AI deployment. The certificate also serves as a stepping stone to advanced ISACA AI credentials such as the AAIA (Advanced in AI Audit) and AAISM (Advanced in AI Security Management).

Certified professionals consistently earn salary premiums over non-certified peers. ISACA's research indicates that certified IT professionals earn an average of 15% more, and certified auditors can command 10–20% higher compensation than non-certified counterparts according to industry salary guides. ISACA's broader credentialing community of 185,000+ professionals spans more than 190 countries, providing global recognition and networking value for this foundational AI credential.

Sample Questions

Preview — answers shown

5 sample questions with correct answers and explanations. Start a practice session to test yourself across all 600 questions.

1. An insurance company is implementing an AI system that will automatically approve or deny insurance applications based on applicant data and risk algorithms, with no human review before decisions are communicated to applicants. Under GDPR Article 22, which statement correctly describes the company's obligations? (Select one!)

AThe system is permitted if it improves processing efficiency and reduces costs for the company

BThe system is allowed as long as applicants are notified that AI is being used

CThe system is acceptable if the AI model achieves at least 95 percent accuracy in testing

DThe system is prohibited unless it falls under specific exceptions and implements required safeguards

Explanation

GDPR Article 22 establishes that individuals have the right not to be subject to decisions based solely on automated processing that produces legal or similarly significant effects. Insurance application decisions clearly fall under this category as they significantly affect individuals. Such systems are prohibited unless they meet one of three specific exceptions: necessary for contract performance, authorized by law with safeguards, or based on explicit consent. Even when exceptions apply, organizations must implement safeguards including the right to human intervention, the ability to express viewpoints, and the right to contest decisions. Simply notifying applicants about AI use does not satisfy Article 22 requirements. Processing efficiency and cost reduction are business benefits but do not override individual rights. Model accuracy alone does not address the fundamental requirement for human oversight in decisions with significant effects. The company must either implement meaningful human review before final decisions or obtain explicit consent while providing all required safeguards including explanation, challenge mechanisms, and human intervention rights.

2. An e-commerce company wants to discover which products are frequently purchased together to optimize product placement and marketing campaigns. They have transaction data but no predefined categories or labels. Which machine learning approach and specific technique should they use? (Select two!)

Multiple correct answers

ASupervised learning with logistic regression

BUnsupervised learning with association rules

CReinforcement learning with Q-learning

DApriori algorithm to identify frequent itemsets

ERandom forest for classification

Explanation

Unsupervised learning with association rules is the correct approach because the company has unlabeled transaction data and wants to discover hidden patterns of co-occurrence. The Apriori algorithm is the specific technique designed to identify frequent itemsets and generate association rules in market basket analysis. Supervised learning requires labeled data which is not available. Reinforcement learning is for sequential decision-making with rewards, not pattern discovery. Random forest is a supervised classification algorithm requiring labeled training data. Association rules mining with Apriori discovers relationships like customers who buy product X also buy product Y.

3. An AI research team trains a Generative Adversarial Network (GAN) to create synthetic medical images for data augmentation. The system includes two neural networks: one generates synthetic images from random noise, and another attempts to distinguish generated images from real medical scans. How do these components interact during training to improve generation quality? (Select one!)

ABoth networks cooperate by sharing learned features to jointly improve image quality metrics

BThe discriminator network pre-trains on real images, then transfers learned weights to the generator

CThe networks engage in adversarial competition where the generator tries to fool the discriminator, driving mutual improvement

DThe generator and discriminator alternate between supervised learning on labeled data and unsupervised feature extraction

Explanation

GANs operate through adversarial training where the generator creates synthetic data attempting to fool the discriminator, while the discriminator improves at distinguishing real from fake. This adversarial competition drives both networks to improve—the generator creates increasingly realistic images, and the discriminator becomes more discerning. The networks do not cooperate or share features; they compete. Transfer learning from discriminator to generator is not how GANs function—they train simultaneously through opposition. GANs use unsupervised training without requiring labeled data, and the training is not alternating between supervised and unsupervised modes.

4. According to IEEE Ethically Aligned Design principles, an AI system designed for public health surveillance must respect internationally recognized human rights. Which principle specifically requires designers to prevent harmful applications of their AI systems? (Select one!)

AHuman rights principle requiring respect for international standards

BWell-being principle prioritizing human welfare

CAwareness of misuse principle preventing harmful applications

DTransparency principle disclosing capabilities and limitations

Explanation

The awareness of misuse principle specifically requires designers and developers to anticipate and actively prevent harmful applications of AI systems. This principle places responsibility on creators to consider how their technology could be misused. The human rights principle focuses on respecting established rights frameworks. The well-being principle addresses prioritizing human welfare in design decisions. The transparency principle addresses disclosure of system capabilities and limitations, not prevention of harmful applications.

5. A computer vision system performs object detection to identify and locate multiple vehicles in traffic camera images. The system outputs bounding boxes with coordinates, confidence scores, and vehicle type classifications. Which metric is most appropriate for evaluating this object detection system's performance? (Select one!)

AR-squared measuring proportion of variance explained

BMean Average Precision (mAP) across object classes and IoU thresholds

CF1 Score combining precision and recall for binary classification

DAccuracy calculated as correctly classified pixels divided by total pixels

Explanation

Mean Average Precision is the standard metric for object detection systems, evaluating both localization accuracy through Intersection over Union thresholds and classification accuracy across object classes. Object detection requires assessing whether objects are correctly located and classified, which mAP captures comprehensively. Pixel-wise accuracy is appropriate for semantic segmentation, not object detection with bounding boxes. F1 Score is designed for binary or multi-class classification without localization assessment. R-squared measures regression model fit for continuous predictions, not object detection with discrete bounding boxes and classifications.

One-time access to this exam

Full access to all 600 questions

Or $15/mo for all 201 exams

Detailed explanations

Free preview stays available