Why a short policy beats none
Employees are already using AI tools. A one-page policy tells them what is allowed, what data must never be shared and who is accountable, which prevents the most common mistakes before they happen. Keep it short enough that people will read it.
Eleven sections to cover
| Section | What it should say | Sample wording |
|---|---|---|
| Purpose and scope | Who and what the policy covers | This policy applies to all staff and contractors using AI tools for company work. |
| Approved tools | Which tools are permitted | Use only tools on the approved list; request additions through IT. |
| Data rules | What must never be entered | Do not enter personal data, customer records, credentials or confidential documents into unapproved tools. |
| Human review | Who checks outputs | A person must review AI output before it is shared externally or used in a decision. |
| Accuracy | Duty to verify | Verify facts, figures and citations independently. |
| Transparency | When to disclose AI use | Tell clients or colleagues when AI materially produced the work, where required. |
| Intellectual property | Ownership and copyright care | Do not submit third-party copyrighted material; check licences for generated content. |
| Security and vendors | Due diligence on tools | New tools need a security and data-handling review. |
| High-risk uses | Where extra approval is required | Hiring, credit, medical and legal decisions need approval from the designated owner. |
| Incident reporting | What to do when something goes wrong | Report suspected data exposure or harmful output to the security team immediately. |
| Ownership and review | Who maintains it and how often | The policy owner reviews it every six months. |
Rolling it out
- Have legal, HR and IT review the draft
- Explain it in a short session with real examples, not just an email
- Publish it where staff actually look, such as the intranet
- Collect questions in the first month and update the wording
This is a starting framework, not legal advice. India's Digital Personal Data Protection Act, 2023 applies when personal data is processed, and AI-specific rules are evolving, so check current guidance and take professional advice.
Related: AI governance certifications, agentic AI in cybersecurity and prompting for managers.
Building AI governance?
Talk to an advisor about AI and governance programmes for managers.
Frequently Asked Questions
How long should an AI acceptable-use policy be?
One page is a good target. A short policy that people read beats a long one they ignore.
What is the most important rule?
Do not enter personal, customer or confidential data into tools that have not been approved for it.
Who should own the policy?
A named owner, often in IT, security or legal, with input from HR and business teams.
How often should it be reviewed?
Every six months is sensible given how quickly tools and rules change.