ROLL IT OUT IN A MONTH, THEN KEEP IT CURRENT

Is your AI policy followable?

A policy on paper0 / 14

Start with the first item. Until people have an approved account, every other rule pushes them toward personal ones.

Everyone has a company-managed AI account they can use today
The tool list names account types in three tiers
Data rules say what may never go into any AI tool
Agents and connectors need admin approval first
The whole policy fits on two pages
A named person owns any AI output that leaves the company
Staff had a short session on risks such as hallucination
There's a no-blame way to report AI mistakes
The policy shows its owner and next review date