Design “AI use case & data classification” checklist (MVP)

### Description

We need a simple, mandatory checklist for all AI work in BOD that reflects Digdir’s AI policy.

Policy requires use of konfidensialitetsklasser 1–3 for data.

«Kun informasjon som vurderes til å være klasse lav (1) eller medium (2) kan lastes inn i en KI-løsning…»

Data above class 1 requires risk assessment.

«…hvis dataen du skal bruke i et KI-verktøy ligger høyere enn konfidensialitetsklasse 1, så skal det gjøres en risikovurdering før bruk.»

This checklist will be the core of KITT’s governance: everyone must fill it in before starting an AI experiment or solution.

TARGET: https://baksia.digdir.no/
INSPIRATION: https://ai-act-service-desk.ec.europa.eu/en/eu-ai-act-compliance-checker

### In scope

_No response_

### Out of scope

_No response_

### Additional Information

_No response_

### Tasks

- [ ] Draft a 1–2 page checklist that covers at minimum:

1. Use case description (problem, user, outcome)
2. Where the result is used (internal, external, decision support, automation, etc.)
3. Systems and datasets involved Data classification (1–3)5. Whether personal data is used (yes/no + type)
4. Whether IP/DPIA or other assessments are needed
5. Tool choice (Digdir-licensed vs private / non-approved)

- [ ] Make the checklist usable as:

1. A Markdown template in this rep: TARGET: https://baksia.digdir.no/
- [ ] Add short “helper texts” for each field so non-experts can understand what to fill in.

Definition of Done
- [ ] Checklist exists as a Markdown template in the repo.
- [ ] An integration with a slack channel or outlook sendt to common KITT email/channel if the teams needs help from us
- [ ] At least one example is filled in to show how to use it.