Home / Glossary / AI acceptable use policy
Glossary
AI acceptable use policy: what it is and what to include
An AI acceptable use policy is a written document that tells employees which AI they may use at work, what data they may put into it and what they must not do. It sets the rules in plain words. A policy tells employees the rules, and it does not show the company which AI they run.
By the AIBound team · Updated 17 November 2026
Key takeaways
- An AI acceptable use policy is the written set of rules for how employees may use AI at work.
- It should cover approved tools, data rules, AI agents and what to do when something goes wrong.
- A policy tells employees the rules, and a separate check has to show which AI they run.
Why an AI acceptable use policy matters
Employees already use AI at work. Without a policy, each person decides alone what is safe to paste into a prompt or which agent to install. A written policy gives everyone the same answer and gives managers something to point to.
It is also the base for everything else. Auditors and customers ask to see it. Security rules, training and tool approvals all start from it. A policy that nobody checks against real use turns into a document that describes how the company hoped AI would be used.
What to include in an AI acceptable use policy
A good policy is short enough to read in one sitting. Most cover the same four areas.
- 1 Scope and approved tools: who the policy covers, which AI is approved and how to ask for a new tool.
- 2 Data rules: which data may go into AI and which may not, such as customer records, source code and secrets.
- 3 Agents and add-ons: rules for AI agents, MCP servers, skills and extensions that act or connect to company systems.
- 4 Duties and follow-up: who owns the policy, how breaches are reported and how often it is reviewed.
AI acceptable use policy vs AI policy enforcement
The policy is the rule on paper. Enforcement is the rule applied at the moment an employee or an agent acts. A company needs the first to have the second.
| Question | AI acceptable use policy | AI policy enforcement |
|---|---|---|
| What is it? | A written document for employees | A technical rule applied to AI as it runs |
| What does it do? | Tells employees what is allowed | Allows, asks about or denies each action |
| What does it show the company? | What employees were told | What AI ran and which actions were stopped |
Sources
- NIST AI 600-1: Generative AI Profile · nvlpubs.nist.gov
Sources checked 19 November 2026
Examples
AI acceptable use policy examples
FAQ
Frequently asked questions
What is an AI acceptable use policy?
An AI acceptable use policy is a written document that tells employees how they may use AI at work. It names the approved tools, the data that may and may not go into AI, and the actions that are banned. It also says who owns the policy and how breaches are reported.
What should an AI acceptable use policy include?
Cover four areas. Scope and approved tools, including how to request a new one. Data rules for customer records, source code and secrets. Rules for AI agents, MCP servers, skills and extensions. And duties: who owns the policy, how employees report a mistake, and when the policy is reviewed.
Is an AI acceptable use policy enough on its own?
No. A policy tells employees the rules. It does not show which AI they run or stop an agent from taking a risky action. Employees forget rules or work around them to save time. A company also needs a way to find the AI in use and to enforce the rules.
How do you enforce an AI acceptable use policy?
Start by finding which AI employees run. Then turn each written rule into a technical one. With AIBound, AgentHarness sets rules as plain-English policy and the agent applies them to every action with allow, ask or deny. AIAssure records each decision as evidence for your GRC program.