Skip to main content
An assessment finds out what could go wrong with an AI system and what should be in place. You answer questions about the AI system in a conversation with Altrum’s assistant. Your answers fill in Altrum’s questionnaire, and submitting them finds the risks that apply, the controls that answer them, and the guardrails those controls switch on. Assessments are optional. You can configure guardrails without one. An assessment never changes whether an AI system is active.

​
Assess, then protect

1

Answer the assistant

The assistant asks one question at a time. Attach documents that describe the AI system if you have them. See Run an assessment.
2

Submit

When the assistant has what it needs, it asks whether to submit. Altrum finds the risks and controls, and switches on the guardrails in scope in your guardrail draft, set to Log only.
3

Review the scope

Review the risks, add risks the assessment did not find, and see which guardrails are configured and which need your action. See Assessment results.
4

Configure guardrails

Add the assistant’s suggested settings to your draft, set up the rest, and test. See Guardrail suggestions.
5

Complete the assessment

Complete Assessment makes your saved draft the live guardrail version and closes the assessment.

​
Who can run an assessment

Owners, admins and the AI system’s owner can start an assessment, answer the assistant, change the risks and complete it. Every member can read the results. One person works on an assessment at a time. When someone else has it open, you see their name and can read but not change it. If they leave without closing it, it becomes free again after a few minutes.

​
Statuses

An AI system has at most one open assessment.

​
Re-assess

When the AI system changes, run its assessment again: on the AI system’s page, choose Start new assessment. Earlier assessments stay in Assessment history with their results and conversations.