A third of your staff already use AI at work, and hide it. Don't ban it, channel it

Here is a number worth sitting with. According to research covered by SMB Tech, 34% of Australian workers use AI tools on the job without their employer knowing, and 44% of businesses have staff using personal AI accounts at work. The Employment Hero survey behind it also found that 42% of workers feel using AI is a bit like “cheating”, which is exactly why they keep it quiet.

So this is almost certainly already happening inside your business. The only real question is whether it is happening with your blessing and some guardrails, or in the dark.

For context on how normal this now is: Australia recently became the heaviest user of Anthropic’s Claude in the world. Your team, and your customers, are already living in these tools.

Why banning it is the worst option

The instinct for a lot of owners is to lock it down. That is the one response guaranteed to give you the downside without the upside.

The productivity is real, so people will not actually stop. In the same research, 75% of workers said AI made them more productive and 74% said it improved their work quality. Tell people to stop doing something that makes their day easier and they do not stop, they just stop telling you. You end up with all of the risk (staff quietly pasting who-knows-what into who-knows-which tool) and none of the benefit (nobody shares what is working, so the good shortcuts never spread).

It gets worse. More than half of workers, 51%, say they learn AI from social media like TikTok and YouTube. If you are not guiding how your team uses these tools, someone with a ring light is. That is ungoverned and badly taught at the same time.

The two risks you do need to manage

Getting out of the way is not the same as looking away. There are two genuine risks, and both are manageable once things are out in the open.

Data going into the wrong place. Staff pasting client records, financials or employee details into a free public chatbot is the real exposure. Accountant Natalie Lennon offers a clean test in SmartCompany: before you paste something in, ask “would I be comfortable emailing this document to a complete stranger?” If not, it does not go into a public tool.

Trusting it too far. As Lennon puts it, AI “often delivers answers with a high level of confidence, even when those answers are incomplete, outdated, or simply wrong”. It is brilliant for drafting an email, summarising a long document or kicking around ideas. It is not your accountant, your lawyer or your strategist. AI is a tool inside a system, not the person in charge of the decision, which is the same point we made about putting AI to work in your actual workflows.

The smarter response: bring it into the light

1. Say it out loud: using AI is encouraged, not cheating. The moment it stops being a guilty secret, the workflows that actually help start getting shared instead of hidden.
2. Give people a sanctioned tool. A proper business AI account with real data protections beats a dozen personal free logins. It costs a little and removes most of the data risk in one move.
3. Write a one-page playbook. What is fine (draft this, summarise that), what is off-limits (client financials in public tools, AI as your tax adviser), and what always needs a human check. One page, plain English, done.
4. Collect the wins. When someone finds a prompt or a routine that saves an hour, capture it and share it. One person’s shortcut becomes everyone’s.

This is about a more capable team, not a smaller one

Worth naming, because 45% of Australian workers, the highest of any market surveyed, fear AI makes them replaceable. The data points the other way: businesses with high AI integration were hiring entry-level staff at twice the rate of low-adoption firms, and 61% of workers said AI was helping them build valuable skills. Handled well, this makes your people better at their jobs, which is the same message as AI is not coming for your team, the admin is.

The func.digital take

Your staff have quietly run an AI pilot for you, for free, and told you nothing. That is a gift, if you take it the right way. The job now is not to shut it down or to wave it through. It is to make the good part official and the risky part safe: sanctioned tools, a one-page policy, and the handful of workflows worth standardising across the team.

If you would like help doing exactly that, setting up tools your team can use safely, a plain policy people will actually follow, and turning the best of the shadow AI already happening into proper systems, that is what a free digital systems audit is for. Get in touch, and turn the guilty secret into an advantage.