[Company]
AI Acceptable Use Policy
- Effective
- [Date]
- Owner
- [Policy owner, e.g. Operations Manager]
- Report to
- [it@company.com]
- Review
- Quarterly
1. Purpose and scope
[Company] uses artificial intelligence (AI) tools to work faster and better. This policy says which tools are approved, what information may be used with them, which decisions always stay with a person, and what is never allowed. It applies to all employees, contractors and anyone using [Company] systems or data.
2. Definitions
- AI tool: any service that generates text, images, audio, video or code, or that summarizes, searches or acts on information, including assistants built into other software.
- Approved tool: an AI tool listed in section 3, used through the company account and plan listed there.
- Agent or connector: a feature that lets an AI tool read from or act in another system, such as email, files, calendars or business applications.
3. Approved tools
No AI tools are approved yet. Until one is, do not use AI tools with company information.
To request a new AI tool or feature, contact [it@company.com]. [Policy owner, e.g. Operations Manager] decides with IT, after checking the vendor's data handling, security and contract terms. Requests are answered within one week.
4. Accounts and access
- Use approved tools only through your company account, signed in with your work credentials and multi-factor authentication.
- Never use a personal AI account, free or paid, for company work or company information, even if it uses your work email address.
- Do not share AI accounts or let others use your session.
- When you leave, your access ends and your conversations stay with the company account.
5. Information levels
| Level | Examples | Where it may go |
|---|---|---|
| Public | Published website content, press releases, public documentation | Any approved tool |
| Internal | Processes, drafts, meeting notes without sensitive details | Nowhere yet |
| Confidential | Customer information, contracts, pricing, financials, employee information | Not in any AI tool |
| Regulated | government IDs | Not in any AI tool |
6. Never share
These never go into any AI tool, approved or not, whatever the account:
- Passwords, API keys, access tokens and other credentials.
7. Acceptable use
- Drafting, summarizing, brainstorming, research and analysis within your role.
- Share only the minimum information the task needs; remove names and identifiers when they aren't necessary.
- Treat AI output as a first draft from a confident assistant that can be wrong.
8. Prohibited use
- Putting information into a tool that isn't approved, or above the level that tool is approved for.
- Creating content that is unlawful, harassing, discriminatory or deceptive.
- Presenting AI output as verified fact, legal advice or a final decision without the review in section 10.
- Creating or sharing images, audio or video that imitate a real person's face or voice without their written consent.
9. High-risk uses: a person decides
AI may help prepare these, but a named person reviews and makes the final decision:
- Decisions about people, such as hiring, pay or discipline.
- Decisions about money, such as payments or credit.
- Anything sent to customers, regulators or the public.
10. Human review and accuracy
- A person checks AI output for accuracy, tone and confidentiality before it goes to a customer, partner, regulator or the public, or into a contract.
- You are responsible for anything you send or publish, whether or not AI helped write it.
- AI-generated code and scripts are reviewed and tested before they run in production or on client systems.
- When content customers rely on is substantially AI-generated, say so where that's expected.
11. Agents, connectors and extensions
- Do not connect an AI tool to email, files, calendars, chat or business systems, or build an AI agent, without written IT approval and a named owner.
- Agents may not send, delete, share or pay without a person approving the action.
- Do not install AI browser extensions, plug-ins or desktop AI apps without IT approval.
- AI meeting recorders and note-takers are used only when every attendee has been told, and not in meetings about regulated information unless the tool is approved for it.
12. Reporting mistakes and incidents
If information went into the wrong AI tool, an AI output caused harm, or an AI tool behaved unexpectedly, tell [it@company.com] right away, ideally within one hour. Reporting your own mistake in good faith will not be held against you.
13. Responsible use
We use AI to support people, not replace their judgment. We are open about where AI is used, we protect the people whose information we hold, and we check AI output for bias and error before relying on it.
14. Training
Everyone completes security awareness training that covers safe AI use and AI-enabled attacks, such as convincing phishing, cloned voices and deepfake video, when they join and at least once a year. Completion is recorded.
15. Ownership and review
[Policy owner, e.g. Operations Manager] owns this policy and reviews it quarterly, and whenever a major AI tool or feature changes. Breaking this policy may lead to disciplinary action.
Acknowledgment
I have read and will follow the [Company] AI Acceptable Use Policy.
Name:
Signature:
Date:
Template from the free AI policy builder at dryruncyber.com/ai-security. Not legal advice.