Why a pass means something

The policy defines 6 roles and requires conflicts to be declared. Appointments to the Examination Board and Impartiality Committee are not complete; the descriptions below distinguish their required duties from completed work.

Academic Board

Reviews the Standard and appoints the Examination Board. Six named members; never writes questions or marks work. Meet the Academic Board

Examination Board

Membership is not complete. The policy assigns this Board approval of questions, pass marks and marking configurations.

Examiners

AI examiners read the evidence and mark your result. A named person countersigns Applied AI Specialist and Applied AI Strategist, which you hold for life. On the paid exams, a person reads every safety event, every cheating flag and every result on the pass mark before release. On the free exam, a person reviews every result before you see it. A person decides every review, appeal and complaint. You can ask for a person to look again at any result.

Registrar

Issues, renews and withdraws credentials, checks identity, runs the public register and handles refunds.

Accommodations coordinator

Decides requests for extra time or other adjustments. Reports to the Examination Board, never to sales.

Impartiality Committee

Membership is not complete. The policy requires a mostly independent committee to review conflicts, exceptions and appeals at least twice a year.

We sit inside Straits AI Group, which also publishes the AI+ Library. Nobody who writes questions or signs results is paid on sales. No course or book is ever required to pass. Publishing-side experts contribute under the Examination Board's rules and the same conflict declarations as anyone else.

• Applied AI Foundations and Applied AI Certified (L1) are approved provisionally, before their evaluation study. What that means for a result.
• We are not accredited by any body, including ISO/IEC 17024. We say so and publish our evidence instead.
• The Examination Board and the Impartiality Committee are not yet fully appointed. We publish the membership when it is complete.
• No study yet shows that a pass predicts how someone works with AI on the job. That study is planned, not done.
• No employer, professional body or government requires this credential.
• Every exam is in English, so it measures English alongside the skill. Other languages are not ready.

The technical manual explains how each figure is made and lists what is still pending.

Reviewer and fairness figures refresh from recorded results each hour. Only exams with a recorded production start qualify; review, sandbox and internal records are excluded. Unstamped history is not included. These rules do not show that the exams are valid. The cumulative ratings below come from a published snapshot.

No reviewer-agreement group meets the publication rule of 30 scored events with a defined QWK. No fairness comparison meets its minimum group size.

The Standard defines capabilities. Each exam version defines its audience, evidence, permitted interpretation, and release status. Crosswalks distinguish broad domain mappings from explicit item evidence; they do not establish equivalent scores.

17 broad mappings; 4 explicit item mappings

US Department of Labor AI Literacy Framework, TEN 07-25 (13 February 2026): 5 content areas, 25 sub-areas

11 broad mappings; 0 explicit item mappings

Anthropic AI Fluency Index (February 2026): the 11 observable behaviours of the 4D framework

7 broad mappings; 0 explicit item mappings

AISA rubric: 11 criteria across 5 dimensions

4 broad mappings; 0 explicit item mappings

Bryq AI Fluency Assessment: 5 dimensions

3 broad mappings; 0 explicit item mappings

Pearson Certiport Generative AI Foundations: 4 domains, 20 objectives

Our practice follows the professional standards for testing. ISO/IEC 17024 covers bodies that certify people; the AERA, APA and NCME Standards cover validity; the ITC guidelines cover online delivery.

The Examination Board sets provisional pass marks. Panels of subject experts replace them and review each mark at least every two years. Every pass mark, result and appeal is written to an append-only log, with who did it and when. A standard that later tightens never cancels a credential already earned.

6 policies govern every credential. These summaries are plain English; the policy text governs where they differ. We drafted them against the 2012 edition of ISO/IEC 17024. The April 2026 edition is not yet reflected.

For anything about a credential, an identity check, an accommodation or a complaint, write to registrar@straitsai.institute.