Responsible AI & Governance
A practical grounding in responsible AI, built on the NIST AI Risk Management Framework, the EU AI Act, and the OECD and UNESCO principles.
Risk tiers: how much governance is enough?
Not every use case deserves the same scrutiny Treating an AI meeting-notes summariser with the same ceremony as an AI credit-scoring system wastes everyone’s time — and treating them the same in the other direction is dangerous. The insight behind the EU AI Act is that governance should scale with risk. The tiering idea Unacceptable […]
Who is accountable when the system is wrong
“The system decided” is not an answer anyone accepts An AI system produces a wrong outcome that affects someone. The question that follows is never technical. It is: who is answerable, and what can the affected person do about it? Organisations that cannot answer that in advance discover, at the worst possible moment, that everyone […]
Assessing a system before you buy it
Most AI in an organisation arrives through procurement Far more AI enters a business inside purchased software than gets built in-house. It arrives as a feature in an HR platform, a scoring tool in a CRM, an assistant bolted onto a helpdesk — and it usually arrives without anyone asking the questions they would ask […]
Incidents: what to do when it goes wrong
Assume it will, and decide now what happens AI incidents are not exotic. A chatbot tells a customer something untrue about their rights. A scoring tool is found to work poorly for one group. A summarisation tool omits a safety warning. Generated content is published containing invented facts. Each of these has happened somewhere, and […]
The questions worth asking every time
Responsible AI is a habit, not a document Most organisations write principles and then struggle to connect them to Tuesday. The connection is a short set of questions asked routinely — in the meeting where someone proposes a tool, not in an annual review. The six Who is affected by this, and were they asked? […]