AI Governance Policy Generator
Pick a few characteristics of your organisation and we'll assemble a draft AI Acceptable Use Policy. Copy or print it, then have your legal counsel review.
AI used in customer-facing workflows, content publishing, decisions with reversal possible
Trade secrets, financials, contracts, customer lists
# AI Acceptable Use Policy — the organisation
_Draft generated 2026-11-18. This is a starting point, not legal advice. Have your legal counsel review before adopting._
## 1. Purpose
the organisation uses artificial-intelligence (AI) tools — including but not limited to large language models, image generators, voice synthesis tools, and AI-powered software — to support its work. This policy sets expectations for how those tools are used, what data may be shared with them, and where human judgement remains required.
## 2. Scope
This policy applies to every person performing work on behalf of the organisation, including employees, contractors, vendors, and volunteers. It covers AI tools accessed through any device — corporate, personal, or third-party — when used for the organisation work.
## 3. Approved tools
Only AI tools on the approved list may be used for work involving non-public information. The approved list is maintained by [TODO: name the team/role] and reviewed every [TODO: monthly / quarterly]. To request addition of a new tool, follow the vendor-evaluation process in section 8.
## 4. Data handling
When entering data into an AI tool, classify it according to the data-classification standard below and follow the corresponding rule:
- **Public:** May be shared with any approved AI tool, including free consumer tiers.
- **Internal:** May be shared with approved enterprise-tier tools only, provided training-on-data is disabled.
- **Confidential:** May be shared only with approved tools that have a signed DPA and confirmed no training-on-data. Audit-log evidence required.
- **Regulated / PII / PHI:** Not currently in scope for AI tool use. Any new use requires approval from the data-protection officer.
Customer data, credentials, source code containing secrets, signed legal documents, financial records, and any information you would not share publicly with the AI vendor must follow Confidential or Regulated handling rules — never Public.
## 5. Prohibited uses
The following uses are prohibited under this policy:
- Generating deceptive content (fake testimonials, impersonation, deepfakes of real people without consent).
- Submitting other people's confidential information (clients, partners, employees) to any AI tool.
- Bypassing safety controls on an AI tool ("jailbreaking") to obtain output the tool would otherwise refuse.
- Using AI to make decisions about hiring, firing, performance evaluation, lending, healthcare access, or legal status without human review and a documented appeals process.
- Presenting AI-generated content as the original work of a named individual where doing so would mislead the reader.
## 6. Human review requirements
Outputs from AI tools must be reviewed by a qualified human before:
- Publication on any external channel.
- Sending to a customer, partner, or regulator.
- Filing with any legal or regulatory body.
- Committing to a production environment if the output is code.
- Making an irreversible business or operational decision.
## 7. Disclosure of AI use
Where AI tools materially contribute to output that is published externally, billed to a client, or submitted to a regulator, that contribution must be disclosed in a manner appropriate to the channel.
## 8. Vendor evaluation
New AI tools must pass the vendor-due-diligence checklist before being added to the approved list. The checklist covers data handling, compliance certifications, access controls, observability, AI-specific risks, and commercial terms. (The published version is here: https://aimodelwiki.com/vendor-due-diligence.)
## 9. Incident response
Suspected violations of this policy, leaked data via AI tools, suspected prompt-injection compromise, or significant AI-tool outages that affect customer-facing work must be reported to [TODO: [email protected]] within 24 hours.
## 10. Training & support
Everyone in scope of this policy must complete the AI-use training course within 30 days of joining or of this policy taking effect, and re-take it annually. Questions about specific uses go to [TODO: [email protected]] or the AI champion in your team.
## 11. Review of this policy
This policy is reviewed every 6 months or when a significant change in tools, vendors, regulation, or organisational risk occurs. Material changes are communicated to everyone in scope.
---
_Generated by the AI Model Wiki Governance Policy Generator. Have legal counsel review before adopting. See https://aimodelwiki.com/ai-governance-policy for the latest version._