|
This newsletter provides practical guidance, tools and resources for the real work of governing safe, secure and lawful AI. |
|
|
Newsletter #54 - March 2026 |
|
|
Stop Copying Frameworks. Start Translating Them. |
|
|
By James Kavanagh
|
|
Why writing your AI governance policy by transposing straight from a regulation, standard or framework is a shortcut to failure. And what to do instead. |
Most organisations approach AI governance the same way. They pick up the EU AI Act, ISO 42001, or the NIST AI RMF and start writing internal policies based on it. They copy the structure. They adopt the language. It feels rigorous. It looks like progress. And it's a reliable path to governance that simply doesn't work.
There's a name for what they're doing: transposition. Copying regulatory language into your own documents without doing the harder work of interpreting what those requirements actually demand for your specific systems, your specific organisation, your specific risk landscape. The article covers why transposition fails, why no single framework can be your policy, and introduces a practical vocabulary for doing the translation properly.
The end goal: a unified control framework where a single internal control can satisfy the EU AI Act, ISO 42001, and a client contract simultaneously. One mechanism doing the work of many, rather than parallel compliance programs quietly drifting apart.
|
|
|
Sign-up for our new FREE Course
Doing the Work of AI Governance covers the adaptive governance approach that underpins everything we teach, through three real case studies of AI governance failure and one of AI governance done right.
|
|
|
|
From Good Intentions to Real AI Governance |
|
|
By James Kavanagh
Our feature article above explains what translation looks like and why transposition fails. This earlier article From Good Intentions to Real AI Governance, asks the question underneath: once you have good controls on paper, what actually makes them work?
It opens with HireVue - a company that had ethical principles, an expert advisory board, external audits and transparency commitments. And still took years to respond to mounting scientific evidence against the very capability at the core of their product. Good intentions, visibly, are not enough.
The article introduces the concept central to Amazon's success where James built their AI Governance framework - the concept of mechanisms. Not policies, not procedures - closed-loop, adaptive systems that make the desired behaviour automatic rather than dependent on teams remembering and executing correctly under pressure. It covers the components every effective mechanism needs, and how individual mechanisms connect into system-level governance that scales. Read the full article
|
|
|
The AI Governance Director's Brief (LinkedIn) |
|
|
Who should lead AI governance in your organisation? |
|
|
By Alexandra El-Shamy (Edition #10: Published March 11, 2026)
|
|
|
Boards and executive teams are asking this question with increasing urgency, and most are approaching it from the wrong direction.
The instinct is to assign it to whoever already owns the nearest thing. Technology? Give it to IT. Regulation? Give it to legal. Product? Give it to product management. Each makes a kind of sense. None are quite right. AI governance draws on multiple disciplines and doesn't sit cleanly inside existing structures.
In Edition 10, Alex works through two questions that should come before the 'who': what role does your organisation actually play in the AI landscape, and what is the real impact of your AI on people and the business? The answers shape what the governance role genuinely demands and where to find the right person to fill it. Read the full article
|
|
|
The AI Governance Practitioner Program
Course 3: Governance Through the AI Lifecycle
US$59
This course takes the governance foundation and policy frameworks you've built and turns them into operating reality. The course includes the seven stages of the AI lifecycle, governance activities at each stage, 35 mechanism cards showing what adaptive governance looks like in practice with defined inputs, outputs, ownership, and feedback loops.
If today's articles raised the question of how to move from governance on paper to governance that actually works, this course is where the answer lies.
|
|
|
|
|
|
As we put the final touches on the last Domain of our AIGP Exam Preparation course, I wanted to highlight the five practice exams we have released, including one that is completely FREE. Now after a couple of weeks of calibration and adjustment on difficulty, I can tell you the average grade is only 71%. It's tough, but that's exactly what you want from a practice exam. If you're thinking about or preparing for your AIGP Certification you can try the full 100-question free practice exam now.
The second thing is big. We're building out a set of tools that will help you understand your (and your organisation's) readiness to do the real work of AI governance, and importantly, where to focus next. More on that soon. We're excited!
As always, if anything in this newsletter or in any of the work you're doing with us has raised questions or sparked something in you, please reply to this email. We read everything and genuinely enjoy hearing from the community.
|
|
|
PS. You're receiving this as a subscriber to communications from AI Career Pro. We respect your privacy, so please unsubscribe through the link below if you do not wish to receive these communications in the future. |
|
|