Someone in your organisation used AI today. You probably did not decide it, and you likely do not know with what data. The responsibility for it still sits with you.
The technology is already in, the governance is not
AI rarely entered the business as an IT project. It arrived in everyday work. Someone pastes a customer email into a free tool to get a quick answer. A department lets an assistant draft proposals. A system you already bought had an AI feature switched on that no one decided to use. None of it passed a decision.
This is not negligence. The benefit is immediate and governance feels like a brake, and in a pressured workday the benefit wins every time. No one chose to skip the control. It simply never reached the table.
So the question is not whether you use AI. You do. The question is who decided how it should be used, and for most the answer is no one.
Three questions you can ask today
Owning your use of AI comes down to three things. You can test them on a single tool right away.
Oversight
Which AI tools are in use across your organisation, by whom and with what data? Most management teams cannot answer, because the tools never passed anyone who would have noted them. Without that picture the rest cannot be governed.
Decision rights
When the AI suggests an answer, who decides whether it holds before it reaches the customer? For a simple draft it is obviously a person. The more you automate, the more often that decision slips to the system without anyone deciding it should. Where the line sits must be settled in advance, not in the moment.
Traceability
If something goes wrong, you are tested on what you can show afterwards. Can you reconstruct what the AI was used for, by whom and on what basis? A tool can do the work. The responsibility for it stays with you.
Take the AI tool you use most and answer the three questions for that one. A single sheet with five columns is enough. The tool, who owns it, what data it reaches, who approves the output, and whether it is written down anywhere. Fill in all five and you have control of that tool. Leave a box empty and you have found a gap, and it took a morning to see.
The regulations have already decided who owns this
This is not a new question. NIS2 places responsibility for risk management and follow-up on the leadership, not on IT and not on the vendor. The EU AI Act sets transparency requirements that apply from 2 August 2026, and the high-risk requirements land in the years after. The need for governance was already there. The rules only put a date on it.
Every use of AI needs the same thing as the rest of the business. Clear ownership, a boundary for what is allowed, and the ability to show afterwards what happened. If it is not decided, it does not exist.
What this means for you
A filled-in list shows where the gaps are. Closing them is another job. Deciding who owns each tool, where the boundaries sit and how you can show it to an auditor, that is governance, and it is rarely done in a morning.
It is also manageable, and it does not slow the pace. Governance is in fact what lets you use AI more widely. Without it, the gap surfaces only once something has already happened, and by then it has become a question from a customer, an owner or a regulator.
The question to take to your next leadership meeting
Who here owns how we use AI today? If no one can answer in the room, you have found your first task.
Want to see how it looks in your organisation?
We start with a review that maps where AI is already in use, where governance is missing and what leadership needs to decide. Get in touch if it becomes relevant this autumn.
Contact us