On 8 September 2026, I had the pleasure of speaking to students on the AI for Business programme at Oxford’s Saïd Business School, presenting on “Explainability & Transparency of AI Models under the EU AI Act.”
With the rapid pace of large AI model development and a string of recent high-profile developments in the field, the topic could not be timelier. At the heart of the session was a question every AI developer now has to grapple with: how do you build genuine trust in an AI model by making it explainable and transparent — both to the people who use it and to the regulators overseeing it?
The session also covered:
- How the EU AI Act’s risk-based framework translates the abstract idea of “transparency” into concrete obligations that differ depending on whether a model is general-purpose or embedded in a specific downstream application.
- The practical tension between protecting proprietary model details and giving users and regulators enough insight to assess a system’s behaviour and risks.
It was a great opportunity to engage with a group of future business leaders on one of the most pressing regulatory and ethical challenges facing the AI industry today.
Published on : 9th of September 2026