Define how AI may help before it can influence important work.

Starter policy structure

Copy the headings below into an internal draft. Assign a Human owner and review the draft with the people responsible for privacy, security, accessibility, records, operations, and legal obligations.

Section 01

Purpose

Name the approved business or mission purposes for AI. Prohibit uses that have not been reviewed.

Section 02

Sensitive information

Identify data that must never be placed into public or unapproved AI systems, including credentials and regulated or confidential records.

Section 03

Approved sources

Require important claims and decisions to trace back to current organizational sources rather than model memory alone.

Section 04

Human ownership

Name who reviews important outputs, who can request corrections, and who holds final decision authority.

Section 05

External action

Separate drafting and analysis from publication, spending, signing, deployment, account changes, and other consequential actions.

Section 06

Evidence

Define what checks, citations, comparisons, or approvals must be recorded before the outcome can be accepted.

Section 07

Incidents

Provide a way to stop work, preserve evidence, notify an accountable owner, and revoke access when behavior is unexpected.

Section 08

Learning

Do not allow drafts or model guesses to become organizational memory automatically. Promote only reviewed learning.

Important boundary

This resource is general educational information. It is not legal, cybersecurity, accessibility, records-management, employment, or regulatory advice; it does not establish compliance.