ChatGPT in Business: The Biggest Risk Isn’t the AI — It’s How Employees Are Using It

South Africa, Durban: There’s a lot of excitement around AI in the workplace right now — and for good reason.

Tools like ChatGPT can help employees write emails faster, structure reports, and save time on admin tasks. In many cases, the productivity gains are real.

But there’s a problem most organisations aren’t paying enough attention to:

The biggest risk isn’t the AI — it’s how people are using it.

What’s Actually Happening Inside Companies

In most businesses, AI adoption isn’t happening through formal rollout plans.

It’s happening quietly.

Employees are already:

Using ChatGPT to draft emails and documents
Asking it to summarise reports
Using it to “sense-check” decisions

And in many cases:

There are no guidelines
No training
No clear boundaries

Not using paid versions with better security!

That’s where the risk starts.

The Part Most People Don’t Understand

AI tools like ChatGPT don’t “know” things.

They don’t check facts.
They don’t verify accuracy.

They generate responses based on patterns — and those responses can sound very convincing.

So what happens in a workplace environment?

An employee gets a confident answer
It looks correct
It gets used in a report, email, or decision

Even if it’s wrong.

AI produces language — not truth.

Industry Perspective

According to Arnold Muscat, Director of College Africa Group:

“Most organisations think the risk sits with the AI itself. In reality, the risk sits with untrained employees using powerful tools without guidance. AI is incredibly useful — but without structure, it can just as easily introduce errors, expose data, or create false confidence in decision-making.”

Where POPIA Comes In

There’s also a compliance angle that can’t be ignored.

If employees are:

Copying client information into prompts
Sharing internal data
Using AI tools without understanding where data goes

…then there’s potential exposure under POPIA.

Not because anyone is doing something malicious —
but because there are no clear rules in place.

The Real Issue: Speed Without Control

AI absolutely improves speed.

But speed without structure creates problems.

Businesses now need to manage:

Faster output
Mixed accuracy
Blurred accountability

Because when something goes wrong:

Who is responsible — the employee or the AI?

What Smart Companies Are Starting to Do

The organisations getting this right aren’t banning AI.

They’re managing it properly.

That includes:

Setting simple, practical guidelines for staff
Training employees on what AI can and can’t do
Making it clear what information should never be shared
Encouraging human review before anything is used

Nothing overly complex — just structured and intentional.

A More Practical Approach

At College Africa Group (CAG), we’re seeing a clear shift in the market.

Companies are moving away from: “How do we use AI?”

To: “How do we use AI safely and properly in our environment?”

That’s where real value sits.

Not in the tool itself —
but in how people apply it.

Enquiries

sales@collegeafricagroup.com
+27 (0) 83 778 4903
www.collegeafricagroup.com

POPIA Note

This solution is designed to remain POPIA-appropriate when implemented correctly. It does not access, share, or store personal data. Organisations should still review their internal policies and compliance frameworks before adoption.

Images:

Most companies aren’t at risk because of AI itself — the real risk lies in how employees are using it without clear guidance, training, or governance.. Photo: Arnold Muscat

Video:

https://www.youtube.com/@collegeafricaexcelvideotip1434