Certified AI Assurance Public Sector Foundation

Who this certification is for

The competency standard for governing and assuring AI used by government and public bodies, where accountability, transparency, and contestability carry particular weight.

Who it is for. Public-sector risk, audit, digital, and procurement professionals, and the suppliers who assure AI for government.

Prerequisites

No formal prerequisites. Candidates are expected to have professional experience in a public-sector, policy, risk, or audit function.

Status

Open for registrationOpen for registration · Examinations from October 2026

This is the Foundation grade of the AI Assurance — Public Sector family — Practitioner and Fellow follow it.

What the examination covers

The examination is mapped to public-sector AI governance and algorithmic accountability frameworks. Weightings are indicative until each syllabus issues.

1. 01

   Public-sector AI use and accountability25%

   Principles for government AI use, and the algorithmic accountability that attaches to administrative decisions.

2. 02

   Transparency and disclosure20%

   Transparency, explainability, and the public disclosure a government deployment is expected to make.

3. 03

   Procurement and supplier assurance20%

   Specifying, evaluating, and contracting for AI so accountability survives the handover to a supplier.

4. 04

   Impact and risk assessment20%

   Impact and risk assessment for public deployments, proportionate to who is affected.

5. 05

   Contestability, redress, and records15%

   Contestability and redress for affected members of the public, and the record-keeping that makes both possible.

Curriculum mapping indicates alignment of learning content with published frameworks. It does not constitute endorsement by, or certification under, any standards organisation or regulator.

What preparation is anchored to

Anchored to recognised public-sector algorithmic-accountability guidance and ISO/IEC 42001 concepts. Candidates prepare independently against the published materials.

The Council certifies individuals — it sells no preparation of any kind, and how a candidate prepared is never visible to an assessor. How impartiality is protected.

One route: the examination

The full rules — registration, identity verification, examination conduct, adjustments, retakes — are in the candidate handbook.

Fees and the pass standard are fixed in the scheme rules and confirmed ahead of the October 2026 examination window.

How candidates are assessed

Examination centred on a supplied public-sector AI deployment, requiring the applicable accountability obligations and the evidence that discharges them.

Every script is marked twice — an AI assessor first and provisional, an independent human assessor second and final. The credential carries a unique identifier any employer can check on the public register.

Questions about this certification

What are the prerequisites?

No formal prerequisites. Candidates are expected to have professional experience in a public-sector, policy, risk, or audit function.

When can I sit the examination?

Registration is open now through the candidate portal; examination sittings run from the October 2026 window and are scheduled at booking.

How is the examination marked?

Every script is marked twice, in a fixed order: an AI assessor first, producing a provisional mark, and an independent human assessor second, whose determination is final on any divergence. No result is issued on automated marking alone, and no adverse decision — a fail, a suspension, a withdrawal — is ever made without a human assessor.

How long is the credential valid?

2 years from the date of issue, recorded on the public register. The term and what follows it are set out in the revalidation policy.

Scheme-wide questions are answered on the FAQ and in the candidate handbook.

Background reading

• What is AI assurance? A practitioner's definition

  Producing credible, independently reviewable evidence that an AI system works as intended — and how that differs from governance and compliance.

• How to build an AI risk profile for a deployment

  A seven-step method for producing a risk profile a sceptical reader can challenge, with the failure mode at each step and what the finished artefact contains.