Your certification guide · Specialist Level
Develop the skills to test AI-based systems, including machine learning and generative AI. Explore AI-specific quality characteristics, data and model risks, and test approaches for systems whose behavior is probabilistic and depends on training data.
The exam uses weighted questions: points are not the same as the number of correct answers. Candidates taking the exam in a non-native language may receive 25% additional time (75 minutes total). Confirm eligibility with your exam provider.
Testers, test engineers, test managers, developers, data analysts, and data scientists involved in AI-based systems.
The ISTQB Certified Tester Foundation Level (CTFL) certificate is required. Prepare through self-study or accredited training, including hands-on work with data and models.
The syllabus at a glance
A concise guide to the certification content in syllabus v2.0.
Introduces AI technologies and how AI-based systems differ from conventional software. Covers narrow, general, and super AI, generative AI, supporting hardware and development frameworks, model hosting, and relevant regulations and standards.
Key topicsAI technologies · Generative AI · Model hosting · Standards
Explains AI-specific quality characteristics, including those in ISO/IEC 25059, and safety considerations. Covers defining acceptance criteria that reflect the behavior, quality, and performance expected of an AI-based system.
Key topicsAI quality characteristics · Safety · Acceptance criteria
Explores forms of machine learning, the development workflow, pretrained models, fine-tuning, and retrieval-augmented generation. Covers data preparation, training, validation and test datasets, classification metrics, neural networks, and neural network coverage.
Key topicsML workflow · Data preparation · Confusion matrices · Neural networks
Explains testing challenges in locked and adaptive systems, statistical test approaches, and difficulties with test oracles. Covers testing generative AI and LLMs, red teaming, exploratory testing, ML-specific test levels, and risk-based testing.
Key topicsTest oracles · Statistical testing · LLM testing · Red teaming
Covers input data risks and how to test for bias, representativeness, dataset constraints, and label correctness. Explains data pipeline testing and practical checks that help prevent unsuitable data from undermining model quality.
Key topicsBias · Data pipelines · Representativeness · Label correctness
Explores model risks, documentation reviews, functional performance testing, and adversarial testing. Covers metamorphic testing when an oracle is difficult to define, drift, overfitting and underfitting, A/B testing, and back-to-back comparisons.
Key topicsMetamorphic testing · Drift · Adversarial testing · Model comparison
Addresses risks across machine learning development and deployment. Explains test approaches and mitigation measures for development activities and checks needed when deploying an ML system into its operational environment.
Key topicsDevelopment risks · Deployment testing · Risk mitigation
Training times are syllabus allocations for accredited courses, not exam section durations or personal study recommendations. Total: 19 hours 30 minutes. Reference material and appendices follow the core chapters in the syllabus.
Study from the source
Check the certification page for exam information, study the syllabus, and use official sample questions and answers to practise.
Official sample answers ↗ · All official downloads ↗
Independent study guide based on the ISTQB CT-AI syllabus v2.0. Syllabus source and copyright: International Software Testing Qualifications Board (ISTQB®) and its authors. This page is not affiliated with or endorsed by ISTQB. Links checked on 7 October 2026.
Ready to apply what you’ve learned?