Insights
AI literacy training for employees: what it must cover, and how to roll it out
AI literacy training for employees teaches staff what generative AI can and cannot do, what data may never be pasted into it, how to check its output, and how to brief it well enough to be useful in their own role. For organisations in Egypt and the Gulf it has moved from optional to necessary for two reasons: staff are already using AI tools whether or not the company has approved them, and firms with operations in the EU now face an explicit AI-literacy obligation under Article 4 of the EU AI Act, applicable since 2 February 2025.
Last updated:
Most organisations in Egypt and the Gulf do not have an AI adoption problem. They have an AI visibility problem. Staff are already pasting documents into free AI tools — on personal phones, on personal accounts, outside any policy. AI literacy training is how a company turns that from a risk into a capability.
What AI literacy training is — and is not
AI literacy training gives every employee a working understanding of generative AI in the context of their own job. It is not a technical course and it does not produce data scientists. It answers five practical questions:
- What is this tool actually doing? Enough understanding to know why it sounds confident when it is wrong.
- What must never go into it? A written boundary for client data, personal data, financials and anything under NDA.
- How do I check the output? Verification habits proportionate to the consequences of being wrong.
- How do I brief it well? A repeatable structure, rather than trial and error.
- Where does it help my work? Role-specific use cases, built on the participant’s own documents.
Awareness sessions usually cover the first question and stop. That is why they change so little.
Why it became urgent
Unapproved use is already happening. Banning AI tools does not stop their use; it only stops the company from seeing it. Training plus a written boundary is the control that actually works.
For companies with EU exposure, it is now an obligation. Article 4 of the EU AI Act requires providers and deployers of AI systems to take measures to ensure a sufficient level of AI literacy among their staff, and it has applied since 2 February 2025. Egyptian and Gulf organisations with EU subsidiaries, clients or operations should check with counsel whether it reaches them.
Local data rules are tightening. Egypt’s Personal Data Protection Law (No. 151 of 2020), Saudi Arabia’s PDPL and Oman’s PDPL all make “what did we paste, and where?” a compliance question. See the data boundaries guide for how to write those rules down.
What a good programme covers
| Module | What participants leave with |
|---|---|
| How generative AI works, without the jargon | A correct mental model: prediction, not lookup |
| The data boundary | A one-page written rule they can apply without asking IT |
| Verification | When to trust, when to check, when not to use AI at all |
| Briefing the model | Two frameworks — POCAB for drafting, GCSE-F for working from documents (explained here) |
| Role-specific practice | Three tasks from their own job, done with AI on anonymised real material |
| Policy and escalation | Who to ask, and what to report |
The last two rows are where literacy turns into use. Training on generic examples produces people who understand AI in principle and never apply it on Monday.
Rollout: who goes first
- Leadership (1 day). The people who approve policy and budget must have used the tools on their own work. Otherwise the first policy question stalls everything. Menova runs this as AI for Executives.
- Champions (2 days). One or two people per department who will field their colleagues’ questions. This is the 16-hour AI for Business workshop.
- Everyone else. Shorter sessions, delivered or supported by the champions, built on the department’s own use cases.
What this looks like in practice
At Kahraba, Egypt’s largest private-sector power generation and distribution company, Menova trained 65 executives across three runs of a 16-hour AI Foundation Workshop, each run of 20 to 23 people. Participants rated it 4.8 out of 5. The full engagement is documented in the Kahraba case study.
How to measure it
Attendance and satisfaction scores show the training happened, not that it worked. Measure three things instead:
- Capability: can each participant complete a real task from their own job with AI, unaided?
- Compliance: is the written data boundary being followed?
- Change: which tasks are done differently thirty days later?
The full method is in measuring the ROI of AI training, and pricing is explained in what corporate AI training costs.
Ahmed Amrousy designs and delivers Menova’s AI programmes. He is Executive Education Instructor at the American University in Cairo’s Onsi Sawiris School of Business and Head of Marketing & PR at Nile Air.