AI vendor risk assessment
One per tool, answered for the plan you’d actually buy.
Tool: [name] · Plan: [plan] · Assessed by: [name] · Date: [date]
Data use
- Does the vendor use our content to train its models on this plan? [Yes / No / Opt-out / Unclear]
- Is that written in the terms or a data agreement we can keep? [Link]
- Do any people at the vendor review our content, and when? [Answer]
Retention and location
- How long is our content kept, and can we shorten that? [Answer]
- Where is it stored? [Country or region]
- Can we delete it, and does deletion cover backups? [Answer]
Access
- If we connect it to email or files, what can it see? [Answer]
- Can it act, or only read? [Answer]
- Does it follow the permissions people already have? [Answer]
Controls
- Do we get single sign-on and central user management? [Answer]
- Is there an audit log we can read? [Answer]
- Can an admin turn features and connectors off? [Answer]
Leaving
- Can we export our content? [Answer]
- What happens to our content when we cancel? [Answer]
Decision
[Approved / Approved with limits / Not approved]. Approved for: [uses]. Not for: [uses]. Review again on: [date].
