Employee AI Usage Policy Template for Malaysian Companies
A plain-language employee AI usage policy template for Malaysian companies: approved tools, do and don't list, PDPA reminders, review rules and a sign-off line.
An employee AI usage policy is the staff-facing version of your AI policy: one page in plain language that lists approved tools, data you must never enter, when to check output and how to report a mistake. The template below can be adapted by any Malaysian company and signed by each employee.
Part of the guide: PDPA and AI in Malaysia: What Companies Must Know (2026 Guide)

On this page
This is the page your employees should actually read. It is the short version of the company AI policy.
Part of the guide: PDPA and AI in Malaysia. Checked 22 October 2026.
What does the template say?
Copy, edit the brackets, and have management and your lawyer approve.
[Company] AI Use: Staff Guide (version [1.0], date [ ])
1. Use only approved tools. Approved: [Tool, company account]. Do not use personal AI accounts for work.
2. Never enter: IC or passport numbers, health information, payroll or salary details, client contracts, passwords and keys, unpublished financial results, anything marked confidential.
3. Customer or staff names and contact details. Enter them only in [tool marked "personal data approved"]. If unsure, replace names with "Customer A".
4. Check before you send. AI makes mistakes and can invent facts. You are responsible for anything you send. Check figures, names, legal or medical statements.
5. Say when AI helped if a customer or the law requires it: [describe cases].
6. Do not use AI to make a final decision about a person's job, pay or credit, to copy others' copyrighted work, or to pretend to be someone.
7. Report problems. If you pasted something you should not have, or an AI output went to a customer wrongly, tell [name/WhatsApp] within [24 hours]. Early reports are welcome and will not be punished.
8. Questions? Ask [name]. We review this guide every [6] months.
I have read and understood this guide. Name / Date / Signature.

Why these clauses?
| Clause | Reason |
|---|---|
| Approved tools only | PDPA Security principle; vendor terms differ by plan (for example, Anthropic treats commercial and consumer products differently) |
| Never-enter list | Sensitive and confidential data need explicit control |
| Check before sending | Data Integrity principle; accuracy is your duty |
| No automated final decisions about people | Fairness and accountability; likely theme of future AI law |
| Report early | Breach notification to the Commissioner can be due within 72 hours |
Why does reporting need a time limit?
The PDPA data breach notification guideline sets 72 hours for notifying the Commissioner of a notifiable breach, and treats a breach affecting more than 1,000 individuals as "significant scale". The clock runs from when the company identifies the breach, so a pasted customer list that a staff member hides for a week costs you the time you needed. That is why clause 7 asks for a report within 24 hours and promises no punishment for early reports. Under the PDPA, health data is sensitive personal data, which is why clause 2 lists it first.

How do you roll it out?
| Step | What to do |
|---|---|
| 1 | Fill in the brackets: tools, named contact, reporting time |
| 2 | Have management and your lawyer approve it |
| 3 | Run the staff session with five examples from your own work |
| 4 | Collect signatures and file them with the staff records |
| 5 | Put the review date in the owner's calendar |
How do you introduce it?
Run a 30-45 minute session with five examples from your own business and let staff ask questions. We run these sessions; see our courses and coaching. HRD Corp levy funding may apply to training delivered by an HRD Corp Registered Training Provider; see the HRD Corp guide.
Participants in Generative AI at Work finish a draft of this policy for their own team in module five.
What next?
Read Shadow AI at work for how to deal with tools people already use, or book a call.
A worked example for accounting practices, with a task table, is in ChatGPT for accountants in Malaysia: safe and unsafe uses.
Sources
(checked November 2026)
- Personal Data Protection Department Malaysia, Act 709
- PDPA guideline on data breach notification
- Anthropic Privacy Center, Is my data used for model training?
The template is general information, not legal advice. Have a lawyer review it, especially clauses on monitoring and discipline.
Frequently asked questions
Should employees sign an AI usage policy?
Yes, a signed acknowledgement shows they read it and gives you a record. Combine it with a short training session so the signature means something.
Can we monitor how staff use AI?
Monitoring staff involves personal data and employment issues. Tell staff in writing what is logged and why, and get HR and legal sign-off before introducing monitoring tools.
Do contractors and interns need to follow it?
Yes. Make the policy part of their engagement terms.