According to this year’s BCG AI Radar survey of 2,360 executives, including 640 CEOs, half of CEOs believe their own job depends on getting AI right. At the same time, nearly three quarters of them say they are the main decision-maker on AI in their company. In other words, they are making decisions they will have to answer for. The question is what they base those decisions on.
Leaders often manage AI blind, yet they carry the greatest responsibility
Organizations have oversight of their finances, security, and people. Not so much of AI.
Overseeing AI is hard because it is starting to touch everything
Companies today have oversight built around every key area of how they operate. Controlling and accounting systems watch over finances. Security teams and their tools watch over IT security. HR systems watch over people.
Each of these layers emerged at the moment when its area became so important and so complex that it could no longer be managed with a meeting and a spreadsheet.
The problem with AI is that it is gradually spreading into every part of the organization at once. It touches finance, security, IT development, HR, and legal compliance, but none of the existing layers covers all of it.
Financial systems see license invoices, but not actual usage and waste. Security tools monitor networks and devices, but not what employees paste into AI chats. HR systems know who has what access, but not who actually uses AI and who ignores it. And compliance teams have no system that continuously documents what AI is doing in the company, even though the AI Act already requires record-keeping and transparency, and from the end of 2027 it will require full documentation, including human oversight, for high-risk systems.
The result is a blind spot right in the middle of the organization. Not because any of these layers has failed. Each one does exactly what it was built to do. None of them was built for this.
Responsibility for AI ends at the top, no matter who you delegate it to
What matters most is who bears responsibility for the consequences of working with AI. Ultimately, it is not the IT department or individual employees. It is the organization’s leadership.
Leadership approves AI budgets and will have to answer for what the company got in return. Leadership will bear the consequences when sensitive data leaks, when AI causes harm to customers, or when the company fails a regulatory inspection. And leadership alone determines whether the company turns AI into a competitive advantage or squanders it, or even lets it create existential problems.
Some organizations respond by appointing AI leaders: Chief AI Officers or AI managers with a mandate to drive AI adoption across the company. But even where an AI leader exists, one thing holds: responsibility cannot be delegated. The board and the leadership team will answer to owners, regulators, and the public regardless of whom they put in charge of the project. That is why they need their own oversight and far better tools that help them, too.
This year’s IBM Institute for Business Value study of two thousand technology leaders in 33 countries shows that this is not a theoretical problem. Two thirds of them say they are responsible for AI systems they do not fully control. Seventy percent add that teams in their company deploy technology faster than IT can keep track of it. If this is the situation of the people closest to the technology, the company’s leadership will not be any better off.
Managing AI well is becoming a measure of management quality
CEOs see it this way themselves. The half of CEOs who, according to BCG, link their position to getting AI right are right about one thing: AI adoption is becoming one of the criteria by which boards, owners, and investors judge the quality of management. Those who get it right will gain an edge. Those who don’t, well, anyone can guess.
And this is the heart of the problem: leadership bears full responsibility for something it cannot see today. It manages AI blind, with no data, no visibility, and no tools. There is no precedent for this in modern corporate management. No CEO would accept managing finances without financial statements or security without monitoring. With AI, many of them do exactly that today.
What to do about it
There is a gap between what leadership knows about AI in the company today and what it will be held accountable for. Appointing someone responsible for AI will not close it. That is a necessary step, but not a sufficient one. Responsibility stays at the top.
The only way to close the gap is to give leadership the same kind of visibility into AI that it has into finances and security. Not deeper technical details. An overview that makes sense to the person who signs the budget and will bear the consequences.
In practice, that means answering four questions that few can answer today: which AI tools are used in the company, who uses them, what data flows into them, and how much it all costs. Plus one more that a regulator will ask a year from now: whether you have all of it documented.
If leadership cannot answer these questions, it is not managing AI. It is just paying for it.
How to get there
The good news is that the answers to these questions can already be found today. There are tools that can connect to existing systems, start tracking AI activity across the entire organization, and give leadership exactly the information it lacks today. We are building one of them ourselves.
See what our tool Asysta Board can do, or go straight to booking a no-obligation consultation. We’ll go through how your organization works with AI and what exactly Asysta Board can do for you.
