BITS Group
Using AI Safely at Work: A Simple Policy for Small Teams

Using AI Safely at Work: A Simple Policy for Small Teams

Small teams across the UK are testing tools like Copilot and ChatGPT to get work done faster. After all, they’re handy for drafting emails or summarising notes. But without clear guidance on what’s safe to share and how these tools fit with data protection, a quick experiment can cause confusion or a data mishap.

Why AI policy matters for small teams

We’re often asked if it’s ok to drop a client file into ChatGPT, or whether Copilot can pull text from customer notes to draft an email. If there’s no AI policy, staff are left guessing. That’s a problem, especially with strict UK data protection laws to consider.

Many small businesses and charities want to get real value from AI while staying on the right side of privacy rules, and feel some concern about which uses are sensible versus those that cross a line. With so much talk about AI, it’s easy to feel caught between ‘move fast’ pressure and the need to protect your data.

Clear rules for powerful tools

We see Microsoft’s Copilot landing in more workplaces and ChatGPT regularly turning up in discussions about efficiency. The UK Government’s Cyber Security Breaches Survey 2024 points out that two-thirds of businesses have cyber security near the top of their agenda, with growing interest in how AI may affect risk.

AI tools often store what you type in, and sometimes use it to improve future responses. If you paste confidential material into these tools, you might expose client data or break data protection laws. The Information Commissioner’s Office (ICO) has shared guidance on AI and data protection, warning not to share sensitive or identifiable information with public AI tools.

Banning AI isn’t the answer. The best way forward is simple, clear rules that put people and data safety first.

What a practical AI policy looks like

An AI policy for a small business doesn’t need to be lengthy or use technical language. It should set out what’s expected and be easy for your team to follow. From what we’ve seen at BITS, the most effective policies:

  • Specify which AI tools are allowed and what they’re for (for example, use to draft generic emails but not sensitive HR templates)
  • Clarify what data must remain private, including client details, commercial secrets or anything identifying someone
  • State who staff can turn to with questions, and encourage them to check before sharing

If you’ve started with Microsoft Copilot, it’s possible to configure which parts of your Microsoft 365 data it can work with. Still, regular staff training goes a long way, in our experience, most mistakes come from not realising what counts as confidential when using AI.

Creating a culture of safe, confident AI use

Rules alone won’t get you the results you need. People are more likely to follow a policy if they understand the reasons and know they can ask questions without hassle. The best AI safety guides are updated as tools evolve. Regular check-ins keep policy fresh and help catch issues early.

We’re big fans of including AI safety in team training sessions, sometimes this is just a five-minute refresher. It’s practical to use examples tailored to your work, like reminding staff that if they wouldn’t send certain data over an unsecured email, they shouldn’t paste it into ChatGPT either.

Your IT provider should also keep an eye on new AI features within existing apps and turn off any that staff don’t need, lowering the risk of accidental over-sharing.

How to build (and maintain) your AI policy

Here’s our tried-and-tested approach for small UK teams after a clear, workable policy:

  1. List which AI tools are allowed for staff. Name them specifically, like Copilot or ChatGPT, and include which versions (such as business rather than personal accounts).
  2. Define what data staff can’t put into AI tools. Keep it plain. No personal data, business-sensitive details or client records. Concrete examples help.
  3. Explain acceptable use of each tool. For instance, using ChatGPT for marketing blurbs is fine; sharing unpublished product details is not.
  4. Set out who staff can ask about AI at work. This could be your IT team, data protection lead or office manager.
  5. Plan regular reviews and training. Aim for at least a quarterly check-in, or when new tools crop up.

The next step: review your team’s approach

Having a clear, adaptable AI policy means you’ll get the benefits from tools like Copilot and ChatGPT while avoiding unpleasant surprises. If you’re not quite sure where you stand or want a calm second opinion, the BITS team offers a free 360 IT Audit or consultation. There’s no pressure, just honest advice on what works for organisations like yours. Give us a call on 0203 011 1190 if you’d like a friendly chat.

Need IT Support?

Contact us today for a free consultation and discover how we can help your business.

Get Started Today