Assurance
Evidence on Your AI Controls
We review your AI controls and test the models you use or build, with findings for your teams and board to review.
Why It Matters
Much AI governance work stops at a policy. A board needs to know whether the controls behind it work. We test them the way an internal auditor tests any other control: what can go wrong, what should stop it, and whether it does.
How We Work
We start from a published method. Our AI Risk and Control Matrix sets out 92 risks across 14 domains, each with a suggested control, the evidence to ask for and tests of design and operation. It is free under the Apache 2.0 licence, so you can see how we assess AI risks and controls before we start.