Assurance

Controls

A Control is a practical, recurring measure that meets requirements or reduces risk. You build it, test it, and keep evidence that it works.

In the app
Project → Understand & Assure → Controls

Example

Requirement: high-risk AI systems need effective human oversight. Control: every high-impact AI recommendation is reviewed by a person before it is acted on.

Control lifecycle

  1. 1Choose New Control, or add one from the Control Library, and link the requirements it covers.
  2. 2Assign an owner and build it. The control moves through Build, Ready to test, Test and Passed.
  3. 3Choose Request evidence or Add evidence to collect proof.
  4. 4Choose Run control test and record Pass, Fail or Not applicable, with the evidence you reviewed.
  5. 5A Fail raises a linked Finding automatically. Choose Remediate to start a corrective action, then Retest when it is fixed.
  6. 6A control can also start from a policy: Create control from this policy adds it to the project's Internal controls and maps it to that policy. Controls you write yourself belong to your project only; they are never offered in the Control Library to other projects.

What a control is not

  • It is not the requirement; it is how you meet it.
  • It is not the one-off task used to build it.
  • It is not the test used to check it.

Last reviewed 27 October 2026

Next stepAssurance methodsDifferent questions need different methods. Several methods can support the same decision.