top of page
What’s inside
- Approved tools. A table of each tool, what it is approved for, the highest data tier it may see, and its owner.
- What you may and must not do. Plain rules, including no credentials in any AI tool and no AI-made decisions about people outside an approved process.
- Human review. You own what you ship: check facts, sources and code before AI output leaves your hands.
- Disclosure and training. When to tell people AI was used, and the AI literacy training the EU AI Act asks deployers to support.
- Requests and incidents. How to ask for a new tool, and a blameless route for reporting mistakes.
How to use it
- Decide the approved tools and data tiers first; most of the policy follows from them.
- Replace every bracketed part and delete any clause nobody will enforce.
- Pair it with the AI Data Handling Policy, which defines the data tiers.
- Walk one team through it before rolling it out.
bottom of page