The Alan Turing Institute recently published one of the best-documented studies on AI governance in business. The central finding is not technical. It is the absence of a framework.
What the Evidence Actually Shows
The two biggest obstacles to AI adoption are the inability to define use cases correctly and the lack of structured risk management. This is not speculation. It is a finding from workshops with 190 organizations and analysis of 40 real AI applications.
At the same time, two pieces of legislation are already changing the rules of the game for every Greek company and organization. DORA has been in effect since January 2025, and the EU AI Act is being implemented in stages, with deadlines that have already begun.
Every automated decision that cannot be explained and documented is an open risk, regulatory, legal, and commercial. And yet, most organizations are still running pilots without governance, without an audit trail, without a clear plan.
What Proper Design Looks Like in Practice
The research describes it in four steps. It starts with mapping, what exactly the AI system does, in which process, on what data, and under which regulatory framework. Then comes diagnosis, where it can fail, and who is affected and how. Mitigation follows, the technical controls, oversight, transparency, and training of people. And finally monitoring, because AI changes its behaviour over time. Governance is not a one-off. It is a continuous process.
Why Agentic AI Raises the Stakes
The arrival of Agentic AI changes the terms entirely. We are no longer talking about tools that answer questions. We are talking about systems that take action, make decisions, and carry out processes autonomously.
This is not something you test in a pilot and see what happens. It is infrastructure. And every piece of infrastructure needs to be designed before it goes live, not after.
The question is no longer whether your organization will adopt AI. It is whether you will do it with a framework, or face the cost of its absence later.
Based on the report “AI Governance for Business: Scoping AI Use Cases and Managing Risks”, Alan Turing Institute / Innovate UK BridgeAI, 2026.
