Compliance guide

AI Acceptable Use Policy Template

An AI acceptable use policy sets the rules for how your business and its people use AI tools like ChatGPT, Copilot and Gemini, safely, securely and responsibly. With customers and security reviews increasingly asking "do you have an AI policy?", this guide gives you a free template, sample wording, and a path to a portal-ready version.

Last updated

Key takeaways

  • An AI acceptable use policy (AI AUP) tells your people which AI tools they can use, for what, and what's off-limits.
  • Its biggest job is preventing confidential, personal or customer data from being pasted into public AI tools.
  • It's fast becoming a standard ask in vendor security reviews, procurement and tenders, having one is a competitive advantage.
  • Emerging frameworks to reference include the EU AI Act, ISO/IEC 42001 (AI management systems) and the NIST AI Risk Management Framework.
  • Write your own from the template below, or have a tailored, portal-ready document done for you.

What is an AI acceptable use policy?

An AI acceptable use policy (sometimes called an AI use policy or AI acceptable use statement) is an internal document that sets out how your people may, and may not, use artificial intelligence tools in their work. It covers approved tools, acceptable and prohibited uses, data protection, accuracy, and who is responsible.

It's the AI-specific cousin of a general acceptable use or IT policy, written for the unique risks of generative AI: data leakage, inaccurate or fabricated outputs, intellectual property, and bias.

Use this as a practical template, not legal advice. Obligations under the EU AI Act or your local privacy laws are worth checking with a professional.

Why your business needs one

  • Data protection, the single biggest risk is staff pasting confidential, personal or customer data into public AI tools. A policy draws the line clearly.
  • Procurement and security reviews, customers increasingly ask suppliers whether they have an AI use policy before sharing data or awarding work.
  • Accuracy and accountability, AI can produce confident, wrong answers; a policy requires human review before outputs are relied on.
  • Consistency, it gives every worker the same rules instead of everyone guessing what's allowed.

Allowed vs prohibited: quick examples

Concrete examples land with a team far better than abstract rules. Adapt this to your tools:

TaskPublic AI toolApproved / enterprise tool
Draft a blog post or generic emailFine (no sensitive data)Fine
Summarise a public documentFineFine
Summarise a customer contractNot allowedAllowed if approved
Enter personal or customer dataNeverOnly if contractually permitted
Generate code for internal toolsReview before useReview before use
Make a hiring or credit decisionNever without human reviewNever without human review

What to include: AI acceptable use policy template structure

Adapt this outline to your business. Keep it practical so people actually follow it:

  1. Purpose and scope, why the policy exists and who it applies to (staff, contractors), and which tools it covers.
  2. Approved tools, which AI tools are permitted and how new tools get approved.
  3. Acceptable uses, the kinds of tasks AI may be used for (drafting, summarising, brainstorming).
  4. Prohibited uses, never enter confidential, personal or customer data into public AI; no use for decisions that need human judgement without review.
  5. Data protection and confidentiality, rules for handling sensitive information and using enterprise vs public tools.
  6. Accuracy and human oversight, AI output must be checked by a person before it's relied on or published.
  7. Intellectual property, ownership of inputs and outputs, and respecting third-party IP and copyright.
  8. Bias, fairness and ethics, awareness that AI can produce biased or unfair results.
  9. Transparency, when and how to disclose that AI was used.
  10. Responsibilities, who owns the policy and who to ask for guidance.
  11. Breaches, the consequences of misuse.
  12. Training and review, how people are trained and how often the policy is reviewed (AI moves fast).

Sample policy statement

"[Business Name] supports the responsible use of AI tools to improve how we work. Our people may use approved AI tools for permitted tasks, but must never enter confidential, personal or customer information into public AI tools, and must review all AI output before relying on it."
The prohibited-uses and data sections matter most. A reviewer wants to see a clear line on what data can never go into a public AI tool.

Download the editable AI acceptable use policy template

Drop your email below and we'll send the complete AI acceptable use policy template (Word and PDF) for you to brand, tweak and sign.

How to write and roll out your AI policy

A policy only works if people know the rules and the rules keep up with the tools.

  1. 1

    Decide your approved tools

    List the AI tools you allow and whether staff use enterprise or public versions, then set a path to approve new ones.

  2. 2

    Draw the data line

    Be explicit about what information must never be entered into public AI tools.

  3. 3

    Require human review

    State that AI output must be checked by a person before it's published or relied on.

  4. 4

    Cover IP and disclosure

    Set rules on ownership of outputs, respecting copyright, and when to disclose AI use.

  5. 5

    Train your team

    Share the policy at induction and refresh it as tools and risks change.

  6. 6

    Approve and review often

    Have it signed and dated, and review it more frequently than a typical policy, AI changes fast.

Free template vs done-for-you document

Two ways to land your AI policy: adapt the free template yourself, or have one written around your actual tools and ready for a customer security review.

Free templateDone-for-you document
Price£0Fixed fee
Effort from youA couple of hours editingA short intake form
Built around your AI toolsYou list them inDone for you
EU AI Act / NIST referencesYou add themMapped for you
Security-review readyFormat it yourselfSupplied as a clean PDF
If a reviewer pushes backYou fix itWe revise it free

Prefer your AI policy done for you?

Tell us which AI tools your team uses and we'll prepare a tailored, signed-ready AI acceptable use policy built to satisfy customer security reviews.

Requests for the AI acceptable use policy are reviewed and prepared manually, we'll follow up by email.

Frequently asked questions

Is this AI acceptable use policy template free?+
Yes, everything on this page, the outline, the sample statement and the examples, is free to use. Requesting a tailored, review-ready version done for you is the only paid option.
What's the difference between an AI acceptable use policy and a statement?+
Very little. A policy is usually the fuller internal document with rules and responsibilities; a statement is a shorter, often outward-facing summary. The same template covers both needs.
What's the most important part of an AI policy?+
The data rules. The biggest practical risk is staff entering confidential, personal or customer information into public AI tools, so the prohibited-uses and data-protection sections do the heavy lifting.
Should my AI policy mention the EU AI Act or ISO 42001?+
If you operate in or sell into the EU, referencing the EU AI Act shows awareness. ISO/IEC 42001 (AI management systems) and the NIST AI Risk Management Framework are useful references too, you don't need certification to align with their principles.
How often should I review an AI policy?+
More often than most policies. AI tools and risks change quickly, so review it at least every six to twelve months and whenever you adopt a significant new tool.