#521 Shadow AI and the Cross-Functional Leadership Gap: The Hidden Risk Leaders Are Already Facing | Barbara Wittmann and Niels Brabandt
Shadow AI and the Cross-Functional Leadership Gap: The Hidden Risk Leaders Are Already Facing
Article based on the interview of Barbara Wittmann and Niels Brabandt
Artificial Intelligence is no longer arriving. It has already arrived.
Yet while many organisations remain focused on formal AI strategies, approved software procurement, and governance frameworks, another reality is quietly unfolding inside businesses across industries. Employees are already using AI, often without permission, structure, or oversight.
In this week’s leadership podcast and videocast, leadership expert Niels Brabandt interviews Barbara Wittmann on one of the most underestimated organisational risks of our time: Shadow AI and the cross-functional leadership gap it exposes.
The term “Shadow AI” refers to employees independently adopting AI tools to improve their daily work when organisations fail to provide sufficiently practical, accessible, or suitable solutions. Often, this begins with seemingly harmless intentions. Employees seek efficiency. They want to solve problems faster, improve productivity, or reduce repetitive work.
However, as Barbara Wittmann explains, Shadow AI grows rapidly and often invisibly.
When organisations move too slowly, employees simply create their own solutions. They register for ChatGPT, Claude, or other AI platforms using corporate email addresses or private smartphones. Even organisations attempting to block access frequently discover that employees bypass restrictions altogether.
The result is an uncomfortable truth for leadership teams.
Many organisations believe they are cautiously evaluating AI adoption. Meanwhile, employees have already made the decision independently.
One striking example from Barbara Wittmann’s experience illustrates the scale of the issue. While serving in an interim technology leadership role, she investigated AI adoption inside an organisation of 1,500 employees. The finding was remarkable: approximately 1,200 employees had independently registered for ChatGPT accounts linked to the organisation’s domain.
From an innovation perspective, this demonstrates enthusiasm.
From a governance perspective, it raises serious concerns.
Without clear guidance, organisations expose themselves to security risks, intellectual property concerns, compliance failures, and significant operational uncertainty. Employees frequently do not know what information is appropriate to share with AI tools and what should remain strictly confidential.
The challenge becomes particularly acute in sectors with strict regulatory, legal, or data protection requirements.
Yet the leadership response cannot simply be prohibition.
As both Barbara Wittmann and Niels Brabandt argue, traditional top-down approaches are unlikely to succeed. AI is not a technology transformation that leadership teams can simply mandate into existence. Unlike many traditional IT projects, Artificial Intelligence is fundamentally behavioural. People will adopt the tools they perceive as most useful.
This reality exposes a broader cross-functional leadership gap.
Too often, organisations position AI exclusively as an IT responsibility. Yet successful AI implementation requires considerably broader collaboration. Technology expertise alone is insufficient. Organisations must combine operational understanding, business process knowledge, human behaviour, compliance awareness, and leadership alignment.
Barbara Wittmann advocates a highly practical solution: invite the wisdom of the organisation itself.
Rather than imposing systems from above, leaders should identify trusted individuals within departments who understand business problems deeply. These people are rarely the loudest voices. In fact, the most valuable contributors are often the quiet professionals whom colleagues already trust for advice.
Interestingly, the ideal AI champion is not necessarily the most technically sophisticated person in the room.
The most effective champions often combine credibility, process understanding, and organisational trust. When paired with technical AI expertise from IT specialists, organisations begin building something far more powerful: cross-functional AI capability.
The implications for leadership are profound.
AI readiness cannot be achieved solely through software licensing or policy documents. Sustainable success depends upon human conversations, trust, experimentation, psychological safety, and shared ownership.
The organisations that navigate this transition successfully will not necessarily be those with the largest technology budgets. They will be the organisations whose leaders recognise that AI implementation is not simply an IT project. It is an organisational leadership challenge.
Shadow AI is already happening.
The only question remaining is whether leaders choose to ignore it or engage with it intelligently.
For further leadership insights, organisational transformation support, keynote speaking, and expert interviews featuring voices such as Barbara Wittmann, connect with Niels Brabandt.
Niels Brabandt
---
More on this topic in this week's videocast and podcast with Niels Brabandt: Videocast / Apple Podcasts / Spotify
For the videocast’s and podcast’s transcript, read below this article.
Is excellent leadership important to you?
Let's have a chat: NB@NB-Networks.com
Contact: Niels Brabandt on LinkedIn
Website: www.NB-Networks.biz
Podcast and Videocast Transcript
Niels Brabandt
AI, and you probably now say, "Well, that's everywhere, isn't it?" And it absolutely is. It absolutely is everywhere. The question is, what is Shadow AI, and why is there maybe a cross-functional leadership gap in your organization? And we have an expert on the matter with us today, backed by a popular amount yet again. Hello and welcome, Barbara Bittman.
Barbara Wittmann
Thanks for having me back.
Niels Brabandt
Thank you very much for taking the time again. What is Shadow AI? What is your definition of Shadow AI, and what is probably the danger of having it in your organization?
Barbara Wittmann
Yeah, so Shadow AI happens when people are finding their own ways of dealing with their daily tasks to handle. So if an organization is a little slow in getting AI going, people sign up to, you know, ChatGPT and other tools and whatnot, whatever is making their life easy.
Niels Brabandt
Mm-hmm.
Barbara Wittmann
And that grows, like, wildfire. It's really it's really hard to put tabs on it. You know, the only the only way to do this as an IT person would be to put something in the firewall to block certain websites or whatever, but no one is doing that. Well, but they just
Niels Brabandt
go on their private smartphones, and then they can do it from there. So they will want to find ways, one or the other. What they probably do is they go on their smartphones on Claude, ChatGPT, Copilot, do it there, and then send it to themselves via email.
Barbara Wittmann
Exactly, whatever. And you know, usually, especially in Europe companies are very conservative about what is the AI tool that is being used for good reasons, you know?
Niels Brabandt
Yeah.
Barbara Wittmann
But you also have to find a balance of moving with the times and evaluating what actually makes sense.
Niels Brabandt
Mm-hmm.
Barbara Wittmann
So it also is a matter of bringing the people to the table to actually understand what is the challenge they're handling. Because if you roll out, let's say, Copilot because you're a Microsoft shop, and it's like, "Oh yeah, this is the AI that we are using," it may not fit the bill for a bunch of people, or they simply don't know how to use it best.
Niels Brabandt
Mm-hmm.
Barbara Wittmann
That's when they go out and look for other things.
Niels Brabandt
Absolutely. And you have a story of a company with 1,500 people there and a couple of free AI accounts you found there. How many were that?
Barbara Wittmann
1,200. So.
Niels Brabandt
1,200?
Barbara Wittmann
Yes. So I did have a stint where I was stepping in as an interim IT person in a company, and I was curious. I'm like, "Okay, no AI strategy. You know, what are people using?" So you can call up OpenAI and actually inquire how many people with the domain ending of the company actually have registered for a free account. And out of 1,500, there were 1,200. So yay. You know?
Niels Brabandt
Yeah, well, yay on the one side that people like the product. However, when I sit in HR or, let's say, legal, I would have sleepless nights over this, especially when there was no formal training and no strategy in place.
Barbara Wittmann
Oh, absolutely. And, and nothing in place where it's like, "This is what's okay to put in there to ask, and this is not." So it's also it's a huge IP risk at the end of the day.
Barbara Wittmann
You know, it's not just an you know, security risk. It's also an IP risk. So that put me in a pickle because it's like, well, if so many people are on, you know, ChatGPT in that case, I need to hustle and get an enterprise license for that.
Niels Brabandt
Yeah.
Barbara Wittmann
And that is a six-digit amount for that you know, volume of people. So basically, you know, if you are a little too slow as a company, it will catch up with you, and then you're forced to step into something that may have not been your first choice.
Niels Brabandt
Yeah, absolutely. And of course, now my question is, when you now say there needs to be a solution, usually people would say, "Look, it's an IT project. The IT is going to lead through it, and the IT is going to figure it out."
Niels Brabandt
This is not how it's going to happen right here, because when the IT tries to own it, people will say, "Look, whatever you own, I will find one way or the other around your idea to use AI right now, right here in my office."
Niels Brabandt
So how do you actually want to get a grip, want to get hold of that, that things do not glow astray?
Barbara Wittmann
Yeah. So it's inviting people into the conversation. AI clearly is not something that you can push down from the top. And, you know, that's how usually big IT changes are happening. It's like, "Hey, this makes sense, and let's just do it." This is one of the first examples where this will fail massively. So we want to invite the wisdom of the crowd. That's the most important thing.
Barbara Wittmann
And the easy piece to do this is find one spokesperson in each department that you basically dedicate as, "This is your AI champion." And sit at a table with them and understand what are the tools they are using what are potential use cases where they could save a bunch of time, and then figure out what actually makes sense, maybe the tools they're already using. And, you know, tell them, "Hey, you know, no harm, no foul. You can share it here. Nothing is going to happen." You also need to create that space where they are willing to share that.
Barbara Wittmann
Maybe it makes sense to go into this bigger and actually make it make it something that more people are using. But it really comes down to we cannot automate that. You have to have a human connection, a human conversation, and keep sharing what works, what doesn't work.
Niels Brabandt
Absolutely. So my question now is, how do you find? Because let's face it, you will find some people who are exactly the right fit. You will find some people who are maybe not too engaged, but they are the right people, and they understand I need to step up to represent my department. However, you will also find the people who think they are the right pick. However, they are maybe very engaged, but just not the right person. So how do you know that you find the right people who actually can help you in the different departments?
Barbara Wittmann
Yeah, it's usually the quiet people. If somebody is in the department and says, "Me me. Hey, hey, you know, I got it." Yeah, don't. Nah. Yeah. Yeah.
Niels Brabandt
Okay.
Barbara Wittmann
So it usually is the quiet people. And how you find out about those is ask around and say, "Hey, if you have a pressing issue, who do you trust and who do you go to for advice?"
Niels Brabandt
Mm-hmm.
Barbara Wittmann
So that's the people you want to have at the table.
Niels Brabandt
Yeah.
Barbara Wittmann
And they may not be the ones that fiddle too much with tools, but they know where the business issue lies.
Niels Brabandt
Mm-hmm. Absolutely. Would you say these people are opinion leaders in the departments?
Barbara Wittmann
I would call them more of a sleeper. You know, they are not really opinion leaders because, again, they don't really take a big stand, have a big voice. But they are trusted. They may be in the in the organization for a fairly long time. They are clearly not, you know, IT geeks in any way, but they understand processes. And that is that is the core here.
Barbara Wittmann
Of course, you want to have somebody from the IT department who's heavily into AI, who's super excited, who knows all the tools. So if you match the the crowd who understands business and one who understands the AI world, that's when it gets interesting.
Niels Brabandt
Brilliant. And when now people say, "Hmm, I think this sounds a bit like us. We did something with AI. Maybe we have Copilot somewhere, but we never actually did anything about it. So maybe we have exactly that issue right now that we are sitting there on a ticking time bomb, and maybe we need help."
Niels Brabandt
And I mean, obviously, you can help here. When people now say, "We'd like to get in touch," how can they get in touch with you?
Barbara Wittmann
Check out my website, digitalwisdom.co. I also have a really interesting calculator tool on there that you can try out to really see if you're not filling these gaps or if you let it sprawl, what is it going to cost you at the end. It's a little bit shocking, the numbers. But hey, try it.
Niels Brabandt
Absolutely. I think these are the perfect final words. So when you say, "I think we have an AI strategy in our head, but maybe not in place, it is now time to act," and Barbara is the person to talk to, so at the end of this podcast and videocast, there's only one thing left for me to say, Barbara, thank you very much for your time.
Barbara Wittmann
Thank you.