AI governance and model policy
Which model providers we use, for what, and how customer data is handled.
Status
This policy is not published yet
It appears here with a version, an effective date, and a plain-language summary once the Head of AI Governance approves it. We do not publish draft policy text.
What this page will cover
- Model providers and their purposes
- Training on customer data
- Retention windows per provider
- Prompt and output logging
- Human review points
- Evaluation cadence and incident classification