74% of Staff Are Already Using AI at Work - Why That Should Worry You

Ask your leaders how many employees use AI today and you will often hear cautious guesses. Ask your employees the same question anonymously and you get a different picture: the majority have already tried AI tools for their work, and many use them regularly.

That gap between what leaders believe and what staff are actually doing is where shadow AI lives - and it is a growing risk.

Shadow AI: the new shadow IT

Shadow AI is what happens when people quietly use unapproved tools to get work done faster. They paste in customer emails, contract wording, internal documents or code snippets, and hope for the best.

Why do they do it? Because they are under pressure, the official tools feel clunky or limited, and the AI on their personal phone or browser tab is just… easier.

The problem is not that employees are curious or proactive. The problem is that, without clear guidance and safe options, they may:

• Share confidential or personal data with consumer tools.
• Rely on AI outputs that contain errors or hidden bias.
• Create unofficial processes that cannot be audited or replicated.

In other words, they are taking real risks on the organisation’s behalf, without being set up to manage those risks well.

Why high usage is both good and bad news

On the one hand, the fact that so many people are already experimenting with AI is encouraging. It means you have a workforce that is open to new ways of working and is actively looking for productivity gains.

On the other hand, uncontrolled experimentation creates three headaches for leaders:

• Risk: you have no clear line of sight on where sensitive data is going or how decisions are being made.
• Inconsistency: some teams are racing ahead while others are left behind, creating uneven service and outcomes.
• Missed value: lots of duplicated effort, with individuals reinventing the wheel instead of sharing what works.

Simply banning AI or pretending it is not happening will not fix any of this. People will keep using whatever helps them hit their targets.

Turning shadow AI into a strategic asset

The answer is not to stamp out AI use; it is to bring it into the light and make it safer, smarter and more aligned with your objectives.

That means:

• Providing secure, approved AI tools that are genuinely useful for day-to-day work.
• Setting clear, simple rules about what data can and cannot be used, and how outputs should be checked.
• Training people in practical skills: good prompting, fact-checking, documenting their AI-assisted work.

An AI academy model is ideal here: instead of a one-off policy announcement, you offer ongoing learning and support that matches how quickly the tech is moving.

How to start the conversation with your teams

If you want honest answers, you need psychological safety. Rather than launching an audit with a threatening tone, try this approach:

• Run anonymous surveys asking how people currently use AI, which tools they rely on and what stops them doing more.
• Host open sessions where staff can share use cases, tips and concerns without fear of punishment.
• Make it clear that the goal is not to “catch people out” but to build better support around what is already happening.

You will likely discover that your people are ahead of your policies - and that they have already found dozens of small, valuable ways to apply AI that you can formalise and scale.

From quiet experiments to confident capability

In a few years’ time, most organisations will look back on this period as the messy middle: the time when employees were experimenting in all directions and leadership was still working out its stance.

You have a choice about how you navigate it:

• Ignore the reality and hope nothing goes wrong.
• Clamp down and drive AI use underground.
• Or acknowledge that your staff are already using AI, and invest in the skills, governance and tools that turn that raw energy into a strategic advantage.

If three-quarters of your workforce are already living in the AI future, it is time your organisation caught up.