SETTING GLOBAL STANDARDS FOR TRUSTED AI CREDENTIALSAI Competence Framework v1.29 · current release
Explainer

What a coverage map claims, and what it does not

Taught and assessed are different claims. The gap between them is where a coverage map is most often used to imply something it does not say.

A coverage map states, for each statement a course or credential references, whether that statement is taught, assessed, or both, with a weighting where it is assessed. That is all it states. It is not a quality judgement, and at Tier 0 it is not a judgement of ours at all.

Taught is a claim about the syllabus

A taught statement means the material covers it. Nobody has established that a learner holding the credential can do it. A course may legitimately teach a great deal that it does not assess — context, background, adjacent practice — and saying so is not a weakness in the map.

An assessed statement means an instrument exists, the learner passed it, and the map says which method was used. That is a claim about the person, not the syllabus, and it is the only kind of claim a credential can properly make.

Where the gap gets exploited

The pattern is a map with eighteen taught statements, three assessed, and marketing copy reading “covers eighteen framework competences”. Every element is true and the impression is false. This is why the word covered does not appear as a value in the schema: the two claims are separated at the level of the field, so that collapsing them requires writing something untrue rather than merely something vague.

The second pattern is exclusions listed without rationale. A statement in a referenced set that the course does not reach must appear with a reason. “Out of scope” is not a reason; it restates the exclusion. Say why it is out of scope, and the map becomes readable by someone deciding whether to buy the course.

What a reader should do with one

Read the assessed rows and the weightings first, and ignore the total count of referenced statements. Then read the exclusions. Then check that the framework version is pinned to a release rather than written as “latest”, because a map against a moving target cannot be verified by anyone, including its author.

What this article discusses

Framework material referenced

These links run one way. The article points at the specification; the specification does not cite the article as guidance.

D6Evaluation and assurance
D6.L2.01Apply defined acceptance criteria to judge whether an AI output meets a stated requirement.
D6.L2.03Document evaluation results so that another person can reproduce the judgement.
D6.L2.06Report evaluation findings, distinguishing measured results from impressions.

Discusses framework version 0.1 · the article itself carries no version

Revision history

No revision since publication

24.08.2026First published.

A notice is never edited. An article may be, and every substantive change appears above with the date it was made.

The author

Ingrid Palm, Registrar

Registrar, AI Certification Standards

Holds the appointed post of Registrar: cuts releases, freezes the dated trees, maintains the deprecation record and administers the register of conformance claims.

Writes the practical guidance on coverage maps and tiers because those are the documents that arrive on this desk incomplete.

Declared interests, in full
AI Certification StandardsDeclared

Employed as Registrar. No economic interest in the organisation, which has no shareholders.

Conformance holdersDeclared

None. Holds no interest in any organisation listed on the register, and none in any applicant.

Cite this article

Ingrid Palm, Registrar (2026) “What a coverage map claims, and what it does not”, Explainer, AI Certification Standards. Non-normative. Available at aicertificationstandards.org/articles/what-a-coverage-map-claims (accessed date).

Cite the article as an article. If you need to cite the competence itself, cite the statement: a credential specification or coverage map should reference statement identifiers, never this page.

Article last updated 24.08.2026 · page last reviewed 30.08.2026 · non-normative, not part of any framework version