Course menu

Module 1: How AI systems work

Welcome: what AI assurance is and how this course works
What AI is (and isn’t)
Data: where models learn what they know
From models to products: generative AI, agents and the AI lifecycle
Module 1 quiz

Module 2: When AI goes wrong: harms and risk

A map of AI harms
How AI systems fail
Thinking in risk
Module 2 quiz

Module 3: Responsible AI principles in practice

Responsible AI principles: where they come from
Fairness in depth
Transparency, accountability and oversight: from principle to requirement
Assignment 1: From principles to requirements
Module 3 quiz

Module 4: AI regulation and standards

The regulatory landscape
The EU AI Act: the essentials
Frameworks and standards: NIST, ISO/IEC and others
AI governance in Africa and data protection
Module 4 quiz

Module 5: Systems thinking and sociotechnical assessment

Why look beyond the model
Mapping the system and its stakeholders
Context, organisations and human–AI interaction
Assignment 2: Sociotechnical analysis
Module 5 quiz

Module 6: Risks of generative and agentic AI

Generative AI: capabilities and characteristic failures
People and generative AI in consumer products
Security risks of LLM applications
Agentic AI: when AI takes actions
Module 6 quiz

Module 7: Evaluating and testing AI systems

What evaluation is for
Measuring performance and fairness
Robustness and evaluating generative AI
Designing an evaluation plan
Assignment 3: An evaluation plan
Module 7 quiz

Module 8: Red teaming AI systems

What red teaming is, and isn’t
Planning and running a red-team exercise
Rating, reporting and acting on findings
Assignment 4: A red-team plan
Module 8 quiz

Module 9: Conducting assurance reviews

The assurance process
Gathering and judging evidence
Impact assessments
Writing findings and driving change
Module 9 quiz

Module 10: Final exam and capstone

Writing an executive summary: a worked example
Final exam
Capstone: an assurance review of “Rafiki”
Where to go from here

Length: About 33 hours, self-paced
Lessons: 49
Language: English

AI systems now decide who gets credit, which patients are seen first and what millions of people are told by chatbots every day. Organisations, regulators and the public need people who can find out, with evidence, whether these systems are safe, fair and lawful. This course trains you to be one of them.

Who it’s for

Anyone who wants to work in AI assurance, governance, policy, audit, risk, compliance, trust and safety, or product, including complete beginners. No technical background or coding is needed: everything is built up from first principles.

What you’ll be able to do

Explain how machine learning, large language models and AI agents work, and where they fail.
Identify and prioritise AI harms and risks using structured methods.
Turn Responsible AI principles into specific, testable requirements.
Apply the EU AI Act, the NIST AI Risk Management Framework, ISO/IEC standards and African data protection and AI policy to real systems.
Analyse AI as a sociotechnical system: people, organisations and context, not just models.
Assess the distinctive risks of generative and agentic AI in consumer products.
Design evaluations and red-team exercises, and rate and report what you find.
Plan and deliver an end-to-end assurance review with findings decision-makers can act on.

How it works

Ten modules, about 33 hours in total, including assignments and the capstone, at your own pace. Each module mixes short readings with interactive practice: scenarios, audits, sorting, risk classification and budget trade-offs. Practice is never marked.

Assessment: a quiz after each module, four applied assignments marked against a published rubric, a timed final exam, and a capstone in which you carry out a complete assurance review of a fictional consumer AI product, with peer review and assessor marking.

Certificate

Complete every step to earn the Certificate in AI Assurance, a verifiable digital certificate and Open Badge. Your capstone result (pass, merit or distinction) is shown on it, and capstone graduates also receive a separate capstone badge.

Certificate and badge

Completed ten modules covering AI fundamentals, harms and risk, Responsible AI principles, AI regulation and standards (including the EU AI Act, NIST AI RMF, ISO/IEC 42001 and African frameworks), sociotechnical assessment, generative and agentic AI risks, evaluation and testing, red teaming and assurance review methods; passed every module quiz, four rubric-marked applied assignments and a timed final exam; and completed an assessor-marked capstone assurance review of a consumer AI product.

Instructor

Kingsley Owadara

AI Assurance: From Foundations to Practice