Two MoTacon attendees are on the left. The MoTaacon logo is in the center, and to the right a prompt to Get Your Ticket.
AI Governance - Whose headache is it anyway? banner image
Chapter

AI Governance - Whose headache is it anyway?

A free event by MoT AI Chapter

I keep coming back to one awkward question whenever AI governance comes up: whose headache is it anyway? Legal, risk, security, engineering, product, compliance? Sometimes the answer is everyone, which can easily become another way of saying no one is quite sure.

What interests me is what happens after the ownership boxes are drawn. A policy gets written. A committee is formed. Controls are defined. Someone says the model has been evaluated and the risks are understood. Then the system meets the real world, where users behave unpredictably, data changes, prompts get manipulated, guardrails fail, and agents start taking actions that nobody fully anticipated.

That is the space I want to explore in this session. If an AI system hallucinates, exposes sensitive information, treats people unfairly, ignores a guardrail, or takes an unintended action, who was supposed to catch it? Who defined acceptable behaviour? Who challenged the assumptions? Who tested the controls? Who decided the evidence was good enough?

The question becomes harder with agentic AI. A chatbot producing a poor answer is one thing. An agent choosing tools, accessing systems, changing records or acting on someone’s behalf is something else. Once AI can act, governance can no longer live comfortably in policies and committees alone.

This is where I believe testing and quality engineering have an important role, but not as the owners of governance or as compliance police. A policy tells us what should happen. Evidence tells us what actually happens. I want to use this session to examine that gap, explore where responsibility really sits, and ask what evidence should exist before anyone claims an AI system is governed.

 

Takeaways

  • Recognise where AI governance responsibility becomes unclear and why that creates real engineering risk.
  • Understand the difference between governing a model and governing the complete AI system around it.
  • See where testing and quality engineering can contribute evidence without becoming the owners of governance.
  • Learn why guardrails, controls and policies need to be tested rather than simply documented.
  • Understand how agentic AI changes the governance conversation once systems can take actions.

----------------------------------------------------------------------------------------

Registrations for our 2026 Chapter Events are live too.

Register for 23 Sept Event: Guardrails & Evals: Keeping Your AI Skills Honest | MoTaverse

Register for 22 Oct Event: The Rise of Agentic Testing | MoTaverse

Register for the 12 Nov Event:  Session with The Thinking Tester (TBC) | MoTaverse

Register for the 21 Jan Event:  If AI is everywhere, what shall I do? | MoTaverse 

 

Thu, 25 Feb 2027
13:30 - 14:30 UTC
Location: Online
Host
Speaker
Subscribe to our newsletter