Employees who use AI without training create five main risks: wrong information, leaked data, uneven quality, hidden rework, and confusion about what is allowed. All five are preventable, and the same short programme prevents most of them.
Wrong information comes first, because AI tools sometimes state errors with full confidence, a problem known as hallucination. An untrained employee copies that error into a report or an email, and from that moment the error carries your company's name.
Leaked data follows closely. Staff paste customer details, contracts or staff records into public tools, and Kenya's Data Protection Act, 2019 sets duties for handling personal data, which is information about identifiable people. A paste into the wrong tool can put you on the wrong side of those duties.
Then the quieter risks. Without shared methods, every employee invents their own way of prompting, so quality depends on the individual rather than on a standard. Unedited AI drafts quietly move work from writing to fixing, so the time saved at the keyboard reappears at the review stage, and managers often miss the swap. And with no policy, cautious staff avoid useful tools while confident staff overstep, which means the company loses on both sides.
Training fixes this cheaply, because one short programme can set your rules and build the habits in the same week.