Why are employees using AI tools without permission - and what does it reveal about your organisation?

(And if you missed the webinar and want the full picture - including what we're calling the two mountains of AI transformation, and why most organisations are still climbing the first one - just drop me a message and I'll happily send across the briefing.)

But I want to dig into Shadow AI specifically. Because I think it's sitting in the blind spot of most organisational AI strategies right now - and the stakes are getting higher.

What do we mean by Shadow AI?

Shadow AI refers to employees using AI tools that haven't been officially sanctioned by their organisation. It's happening everywhere. And the important thing to understand is: it's not happening because people are being reckless or careless. It's happening for deeply human reasons.

Before we get to the "why", it's worth sitting with the "how much." There's compelling research emerging on the cost of what Forrester are calling "shadow work" - the invisible, unofficial tasks that drain employee time and energy outside of their core role.

Now, shadow work and Shadow AI aren't the same thing. But they're deeply connected. AI tools are increasingly what people are reaching for to manage those exact kinds of tasks - and when organisations don't provide them officially, people find their own. The demand is there whether or not the governance has caught up.

In our investigations and conversations with organisations, we've identified four distinct drivers. We call them the four S's.

The first S

Sanction - The Policy Gap

The first driver is, in some ways, the most straightforward to address. People are using Shadow AI because the organisation hasn't been clear. There are fuzzy boundaries around which tools are approved, what data can be shared with an AI model, and when AI is appropriate at all. In the absence of that clarity, people default to "easier to ask forgiveness than permission" - and quietly use the tools that work best for them.

This isn't a technology problem. It's a leadership and communication problem. And it's one that can be addressed relatively quickly when there's will to have the conversation openly.

The second S

Shame - The Psychological Gap

The second driver is more nuanced, and honestly, it's the one I find most interesting right now.

In some organisations, using AI is still quietly seen as cheating. There's a cultural undercurrent that says: if you needed AI to do that, it doesn't quite count. People feel like they're taking a shortcut, and they'd rather keep it in the shadows than risk being seen as less capable or less intelligent.

What's interesting - and what we're starting to see in our research - is that this dynamic is beginning to flip. The shame is starting to work the other way too. People who aren't engaging with AI are beginning to feel left behind. And that's its own kind of pressure.

Both versions create a climate where people hide what they're actually doing. Neither is healthy. What organisations genuinely need is a culture that supports open learning, honest experimentation, and real conversation about where AI helps and where it doesn't. And I know a thing or two about what it takes to build that kind of culture! It doesn't need to be complicated - just consistent. It needs to go way beyond the policy memo.

The third S

Superstar - The Competitive Gap

The third driver is perhaps the most strategically significant for leaders.

When Shadow AI is giving someone a real productivity advantage - helping them work faster, produce stronger outputs, finish hours earlier - they have a genuine incentive to keep it quiet. It's become part of their edge within the organisation. Sharing it would close the gap they've worked to create.

This means organisations can end up with hidden performance inequality they have no visibility over. Some people are quietly supercharged. Others aren't. And neither group is having an honest conversation about it. Getting these people out of the shadows will give you an organisational leg up - learn from what they've implemented.

The fourth S

Superiority - The Capability Gap

The fourth driver is the one that's moving fastest right now.

The AI tools people have access to in their personal lives - through their own accounts, their own subscriptions - are often significantly more capable than what their employer has officially approved. Enterprise versions, particularly in regulated sectors, often get stripped back in ways that limit their usefulness. So people find themselves doing work with a personal tool that the official one simply can't match. And the gap between the two, for many people, isn't narrowing.

So what does this mean for you?

If you're leading an organisation, the question isn't whether Shadow AI is happening. The chances are it is. The question is whether you understand what's actually driving it - and whether your response is going to address those root causes, or just add another layer of policy that creates compliance anxiety without changing anything on the ground.

Over the coming weeks, I'm going to go deeper on each of the four S's - because each one has a different shape, and needs a different kind of response.