AI and Governance

This module sets out why artificial intelligence now belongs on every board, cabinet and trustee agenda rather than with the IT team alone. It separates governing AI from AI in governance, draws on the National Audit Office's 2024 report and director surveys, and uses the Committee on Standards in Public Life's 2020 review of the Nolan Principles, the EU AI Act's literacy duty and six plain board questions.

This module examines what the duties of company directors, charity trustees, councillors, officers and housing board members mean when AI shapes decisions. It covers sections 155, 172 and 174 of the Companies Act 2006, public law, housing regulation and the Charity Commission's April 2024 position, with the Deep Knowledge Ventures VITAL episode and the OpenAI board crisis of November 2023, and closes with a practical accountability map.

This module looks at how boards, cabinets and trustee bodies give real oversight to AI through direction, risk appetite, structure and reporting. It works through the Institute of Directors' twelve principles for AI governance, five building blocks, questions to ask of an AI strategy, a risk appetite spectrum from averse to eager, four places oversight can sit, four board postures and what a good quarterly AI report contains.

This module explores how boards and audit committees obtain real assurance on AI rather than reassurance. It applies the Three Lines Model, the Institute of Internal Auditors' AI Auditing Framework and the NIST AI Risk Management Framework, explains ISO/IEC 42001, the ATRS and the UK AI Playbook, and sets out proportionate assurance, an audit committee agenda informed by the 2026 Audit Wales Code of Audit Practice, and a six-step assurance plan.

This module examines how organisations make AI visible to the people it affects, starting from the Committee on Standards in Public Life's 2020 finding that openness and accountability are under strain. It covers the Freedom of Information Act 2000, the Algorithmic Transparency Recording Standard, a transparency ladder, the Dutch childcare benefits scandal, scrutiny lines of enquiry, automated decisions under the Data (Use and Access) Act 2025 and Welsh routes to worker voice.

This module turns to AI used by board members, councillors, trustees and governance professionals in their own work. It draws on the Diligent Institute's 2026 director survey and the Charity Commission's 2024 position to weigh what AI can add and take away, and sets out safeguards for AI note-taking, the governance professional's role as custodian of the record, an illustrative case and a protocol for members' own use.

This module explores how boards turn stated values into decisions about AI, including uses they rule out. It sets out three competing meanings of fairness from Kleinberg, Mullainathan and Raghavan and from Chouldechova, the OECD AI Principles, ethics panels including West Midlands Police's data ethics committee, five steps from values to decisions, red lines, and the Equality Act 2010 with the Welsh Socio-economic Duty.

This module examines cases where governance failed and people paid the price: Australia's Robodebt scheme, the Post Office Horizon scandal and the 2024 deepfake video call fraud at Arup, with the Dutch Toeslagenaffaire alongside. It draws out five patterns behind algorithmic failure, the questions a board could have asked, six steps for when an AI incident reaches the board and how to make sure bad news arrives early.

This capstone tests the topic's frameworks against two illustrative cases set in Wales: Cyngor Sir Glanmor, an invented council whose AI fraud-scoring proposal is called in by scrutiny, and Ymddiriedolaeth Cwm Aur, an invented young carers' charity where a trustee summarises a safeguarding report with a consumer tool. It reads both through five lenses and closes with a five-layer synthesis, a self-assessment and a board's first ninety days.