Ethics for AI, worked through in the open.
Short, specific notes from real casework: what actually changed a board's decision, what a reporting line needs to work under pressure, and what a defensible AI decision looks like written down.
Ethics by design versus ethics bolted on at the end
Ethics bolted on at the end catches what’s already shipped: a pre-launch review that flags a problem in a system that’s mostly built, mostly funded, and psychologically hard to send…
The moral responsibility gap, in a real incident
When an automated decision leads to real harm, we often lump together three important questions: who caused it, who is morally responsible, and who is legally accountable. But these are…
How we actually decide something is right, not just legal
Our Approach breaks down our work into three clear steps: frame the question, narrow the scope, and deliver results transparently. But what happens inside that first step, framing, is where…
What Article 4 actually requires, and what the market is selling instead
Somewhere right now, someone is finishing an hour of AI-literacy e-learning that will not make them meaningfully more literate about anything. That is not a knock on them, it is…
What actually makes someone qualified to hold this role
The role of an AI ethicist is often misunderstood in hiring, frequently seen as simply a personality type: someone reflective, who is not afraid to ask the tough questions. While…
Four layers of governance, one very lonely layer of accountability
Most AI governance frameworks tend to follow a familiar pattern: the board sets the high-level policies but rarely dives deeper than the executive summary; the audit committee reviews details intermittently;…
Reactive, defensive, accommodating, or leading: what is your AI governance actually doing?
Ask a board how mature its AI governance is and you will get a noun back: a policy, a framework, a slide with a maturity score on it, usually somewhere…
What EN 18274 will actually say, and when to expect it
EN 18274 has been the reference everyone keeps namedropping, sometimes accurately, often not. It has passed its final vote at CEN-CENELEC and is expected to publish sometime in 2026, the…
The escalation route: what “protected disclosure” actually means in practice
An escalation route is a defined path for raising a serious concern about an AI system that does not run through the person who has a reason to suppress it,…
Prefer a feed reader? Subscribe via RSS. Prefer email? Use the form below.
The Ethics Directive
A deeper, annual look at the state of the AI Ethicist role across sectors.
View the report →