The short answer: An AI acceptable use policy is a one-page set of rules for how your staff use tools such as ChatGPT, Claude, Gemini and Copilot at work. It names the approved tools, the data never to paste, the check before anything goes out, and one person who owns it. Copy the free template below and adapt it today.
Your staff are probably using AI already, on their phones, on personal accounts, with your customers' details. A policy does not stop that. It makes it safe. This post gives you the whole template as text, explains each part, and shows how to roll it out in a single staff meeting.
- What it is: a one-page AI policy for employees, in eight short sections
- The eight sections: purpose, approved tools, data never to share, review before use, disclosure, accounts and settings, incidents, owner
- Who it suits: any Kenyan business or team from three people upwards
- Cost: free. Copy it from this page. No email needed.
- Time to roll out: one 45-minute staff meeting
- Review: every six months, or when you add a tool
Why does one page beat ten?#
Because staff read one page and ignore ten. A short AI policy that everyone knows protects you better than a long one that sits in a shared folder.
Long AI policies tend to copy legal language from large foreign companies. They talk about "large language model governance" and "algorithmic accountability". A receptionist in Kisumu or a sales officer in Thika cannot act on that. They need to know three things: which tools they may use, what they must never paste, and who to call when something goes wrong.
The problem is real. In Microsoft and LinkedIn's 2024 Work Trend Index, 75% of knowledge workers said they use AI at work, and 78% of those users bring their own AI tools rather than tools their employer gave them (global survey, published May 2024, checked October 2026). That is the gap a one-page policy closes: it moves staff from private, unchecked use to agreed tools with agreed rules.
A one-page policy also gets updated. When a new tool arrives, the owner changes one line in the approved tools section and tells the team. A 20-page policy is rarely touched again after it is signed.
What goes in an AI acceptable use policy?#
Eight sections, each a few lines long. Together they answer what staff may use, what they must protect, and what happens when something goes wrong.
Here is why each section is there:
- Purpose: says who the policy covers. Contractors and interns paste work into AI too, so name them.
- Approved tools: a short list beats a ban. Staff who are told "no AI" use it anyway, on personal phones, where you cannot see or control it.
- Data never to share: the most important section. It turns the Data Protection Act into a list a new employee can follow on day one.
- Review before use: AI writes fluent text that can be wrong. A wrong figure in a quotation or a made-up clause in a contract is your company's mistake, not the tool's.
- Disclosure: some tenders and clients now ask whether AI was used. Staff need to know when to say so.
- Accounts and settings: company data on a personal account leaves with the employee. A company account stays with the company.
- Incidents: people hide mistakes they fear being punished for. A clear, calm reporting rule gets you the news while you can still act on it.
- Owner: without a named person, the approved tools list goes stale within months.
If your business already has an ICT policy or a confidentiality clause in staff contracts, keep this policy separate and refer to them. A separate page is easier to update when the tools change, and they change often.
The template below is a general version of the AI policy packs we write for teams we train. Every client detail has been removed. Words in square brackets are for you to fill in. Delete any example that does not fit your business.
The free AI acceptable use policy template#
Copy everything from "AI Acceptable Use Policy" down to the signature line into a Word document, fill in the brackets, and print it on your letterhead.
AI Acceptable Use Policy#
[Company name] · Version 1.0 · Approved by [name, role] on [date] · Next review: [date, six months later]
1. Purpose#
This policy lets our staff use AI tools to work faster while protecting our customers, our staff and our business. It applies to everyone who works for [Company name], including part-time staff, interns and contractors, on any device used for work.
2. Approved tools#
You may use these AI tools for work: [for example: Claude on the company account, ChatGPT, Microsoft Copilot with your work login]. Use only the accounts listed here. To add a tool, ask the policy owner first. Do not install AI browser extensions, add-ons or meeting note-takers without approval.
3. Data you must never put into an AI tool#
- National ID numbers, KRA PINs, passport numbers and dates of birth
- M-Pesa statements, bank account numbers and card details
- Customers' or colleagues' names together with their phone numbers, addresses or email addresses
- Health, salary or disciplinary information about any person
- Passwords, PINs and login codes
- Anything a contract or confidentiality agreement says we must keep private, including tender prices and unpublished financial results
Replace real details with labels such as "Customer A" or "[amount]". If you are not sure, ask before you paste.
4. Review before use#
AI tools make mistakes and state them confidently. Before anything made with AI leaves the company, a person reads it in full and checks every name, figure, date, quotation and legal or tax statement against the source. The person who sends it is responsible for it, not the tool.
5. Disclosure#
Tell your manager when AI did most of a piece of work. Tell a customer when they ask, and whenever a contract or tender requires it. Never use AI to create the image, voice or signature of a real person.
6. Accounts and settings#
Use the company account where one exists. Never use a personal AI account for customer or company information. Where a tool lets you stop your chats being used to train its models, switch that off. Do not share logins. When you leave [Company name], your access to company AI accounts ends.
7. Incidents#
If you paste something you should not have, or an AI answer causes a problem, tell [policy owner] the same day. Delete the chat where the tool allows it. Reporting a mistake early is never a disciplinary matter in itself; hiding one is. If personal data is involved, the policy owner decides at once whether the Office of the Data Protection Commissioner must be told.
8. Owner#
[Name, role] owns this policy. They keep the approved tools list up to date, answer questions, and review the policy every six months or whenever we add a tool. Questions go to [email address or WhatsApp number].
I have read this policy and I will follow it. Name: ________ Signature: ________ Date: ________
How do you roll out an AI policy in one meeting?#
Put the policy in front of everyone in a single 45-minute meeting, show one example, and collect signatures before people leave. Sending it by email alone does not work: people sign without reading.
- Five minutes, why. Say plainly that AI is welcome and that the policy exists so people can use it without risk.
- Ten minutes, the eight sections. Read the approved tools and the "never paste" list out loud. These two parts matter most.
- Fifteen minutes, one live example. Take a real task and remove the personal details on screen. For example, at Riziki Beauty Lounge (an invented salon we use in class), "Remind Wanjiru Kamau, 0712 345 678, that her Ksh 4,500 balance is due" becomes "Write a polite reminder to a client that her balance of [amount] is due". The answer is just as good.
- Ten minutes, questions and tools. Ask which AI tools people already use. Add the ones you approve to section 2, today.
- Five minutes, sign and share. Collect signed copies, and pin the policy in the staff WhatsApp group.
Check in after 30 days. Ask what people use AI for, whether any tool needs adding, and whether anyone pasted something they should not have. Change the policy if the answers say so. Our guide on how to train your staff on AI covers what to measure in those first 30 days.
How does an AI policy link to Kenya's Data Protection Act?#
The policy is how you show that your business takes reasonable steps to protect personal data when staff use AI. The Data Protection Act, 2019 makes your business responsible for the personal data of customers and staff, whichever tool processes it.
Three parts of the Act matter most for AI use at work (Act checked on kenyalaw.org, October 2026):
- Data minimisation. Personal data must be limited to what is needed for the purpose. Section 3 of the template, the "never paste" list, puts this into practice.
- Transfers outside Kenya. Part VI of the Act sets conditions for sending personal data out of the country. Most popular AI tools process data on servers abroad, which is one more reason to strip names and numbers before you paste.
- Breaches and penalties. A data controller must notify the Data Commissioner within 72 hours of becoming aware of a breach that poses a real risk of harm. Under section 63, the Commissioner can impose a fine of up to Ksh 5 million or, for a business, up to 1% of the previous year's turnover, whichever is lower.
The Office of the Data Protection Commissioner publishes registration rules, guidance notes and complaint decisions. Read our full guide to AI and Kenya's Data Protection Act for the settings to change in each tool and the five rules we give teams. This post is general information, not legal advice. For a decision about your business, speak to an advocate.
Frequently asked questions#
Does a small business need an AI policy?#
Yes, if anyone on your team uses AI for work. A three-person business still holds customer names, phone numbers and M-Pesa records, and the Data Protection Act applies to it. A one-page policy takes an afternoon to adapt and gives staff clear answers on which tools to use and what never to paste.
Who should own the AI policy?#
One named person, usually the owner, the operations manager or the HR lead. They do not need to be technical. Their job is to keep the approved tools list current, answer staff questions, receive incident reports the same day and review the policy every six months. A policy owned by a committee is a policy nobody updates.
How often should we update our AI policy?#
Review it every six months, and update it straight away whenever you approve a new tool, a tool changes its data settings, or an incident shows a gap. Most updates are a single line in the approved tools section. Tell the team in the staff WhatsApp group and keep the old versions on file.
A policy works best when staff also know how to use the tools well. Cloudwise Limited, an ICT company in Westlands, Nairobi, runs hands-on AI training on Claude for individuals and teams. Our AI training for teams includes this AI use policy template, adapted to your business, alongside hands-on practice on your own documents.
Training a team of 10 or more? Ksh 13,500 per person, built on your own documents. Book a 30-minute call at cloudwise.co.ke/ai-training/teams.
Written by
Gavin



Comments
Be the first to comment.
Leave a comment