AI governance covers the day-to-day responsibilities, decisions, controls and records around AI. AI assurance examines the evidence to see whether a system, process or governance claim meets stated criteria.
At a glance
| Activity | Primary question | Typical output |
|---|---|---|
| AI governance | How do we direct, control and remain accountable for AI? | Ownership, policies, registers, controls, monitoring and records. |
| AI risk assessment | What could go wrong in this use, who is affected and is the remaining risk acceptable? | Use-specific risks, controls, owners and decisions. |
| Advisory governance audit | What does the evidence show about our current governance position and gaps? | Scoped findings, priorities, limitations and recommendations. |
| Independent assurance | Can an intended user rely on a defined claim against stated criteria? | An assurance conclusion or report, with scope and level clearly stated. |
| Accredited certification | Does a management system conform to a certifiable standard under an accredited process? | A certification decision by an appropriately accredited certification body. |
Why the distinction matters
An organisation can have governance without independent assurance: it may operate policies, approvals and monitoring internally. Assurance then examines a defined claim or control environment. Certification is narrower still in meaning and should not be implied by an advisory review.
Independence has levels
A person can be independent of the work they are reviewing without being an accredited certification body. Conversely, a provider that helped design a control may face a conflict when asked to provide assurance over that same control. The intended user, criteria, scope and provider role should all be explicit.
Questions to ask before commissioning work
- What exact claim or question will the work answer?
- What criteria or framework will be used?
- Is this advice, internal audit, external assurance or accredited certification?
- Who will rely on the result?
- What evidence and systems are in scope?
- What conflicts or limitations exist?
- What wording will appear in the final conclusion?
Where Mise sits
Mise helps organisations improve operations with process change, automation and AI, and builds governance into that work: ownership, oversight, controls and evidence. We do not provide accredited certification or independent assurance opinions, and we do not describe our work as a certification outcome.
Read what an AI governance audit should contain or see how we build governance into delivery.