AI Policy
Launch Checklist
Turn a draft into rules your team can find, understand and use. Work through these four stages with the people who own the decisions.
1. Make the rules specific
- Name the policy owner and the teams, contractors and work it covers.
- List approved tools, account types and permitted uses. Say where staff can check the current list.
- Define information that must stay out of AI tools, the review needed before delivery, and limits on connected actions.
Owner:Target date:Record or open question:
2. Review and approve
- Walk through one real workflow with its owner. Check that the rules can be followed in the tools and accounts people actually use.
- Resolve client commitments, information-handling requirements and open questions with the appropriate reviewers.
- Record the approver, version and effective date. Set a clear route for tool requests, exceptions and concerns.
Approver:Effective date:Approved version:
3. Brief the team
- Publish the approved policy in one location. Explain what changed and which workflows are affected.
- Use one permitted example and one prohibited example from your team’s work. Show how to request help or report a concern.
- Record who received the briefing and any acknowledgments your organization requires. Include new starters and relevant contractors.
Briefing owner / date:Policy location:Questions to resolve:
4. Check it works
- Set the first review date and the events that trigger an earlier review, such as a new tool, changed client requirement or incident.
- Check a small sample of actual work for approved account use and recorded human review. Track findings without assuming that a policy proves compliance.
- Assign owners and dates to fixes. Update the rules and brief affected people when a change is approved.
Review owner / date:Work sample:Follow-up record: