The AI Competence Framework
The AI Competence Framework defines what people need to know and be able to do to work effectively with AI. It is organised into eight domains of competence, four levels of proficiency, and a set of overlays and extensions that adapt it to particular professional roles and contexts.
The framework is a specification. It describes competence. It does not prescribe how competence is taught, how it is assessed, or by whom it is recognised. Those decisions belong to training providers, employers and awarding bodies, who apply the framework independently and without charge.
Version 1.29 · issued 23.08.2026 · open for comment · licensed CC BY 4.0
Scope
The framework covers the competence required to use AI systems in professional work: to select them, direct them, integrate them, judge their output, and remain accountable for the result.
It is written for the person applying AI to a task, not for the person building the model.
What the framework does not cover
Published as a list, because scope claims made only in the positive are read as broader than intended.
Observable
Every statement describes something a person does, which another person can observe. Statements beginning “understands” or “is aware of” are not assessable and are not published.
Tool-neutral
No product, vendor or model name appears in any statement. Products change faster than competence does.
Portable across jurisdictions
Where obligations differ by jurisdiction, the statement describes the obligation and the indicators note that the specifics are local.
Additive
New contexts are served by adding extensions, never by modifying core statements. A credential awarded against v1.29 does not become invalid because a new sector was added in v1.30.
Permanently identified
Every statement has an identifier and an address. Neither is reused, renumbered or removed.
Openly licensed
The content is free to use, adapt and build on commercially. Only claims of conformance are controlled.
Six routes in
The framework is read for different purposes. Each route below opens at the page that answers that purpose directly.
| If you | Start at |
|---|---|
| Want to understand how the framework is built | /framework/structure |
| Want to know what a level means | /framework/levels |
| Are looking for a specific competence | /framework/statements |
| Teach or assess AI competence | /using/map-a-course |
| Are writing a role description | /using/employers |
| Are building a credential | /using/build-a-credential |
Where the framework stands
The framework is published as it is written. All eight domains are entered in the statement index at version 1.29.
Published 255 · across 8 of 8 domains, three overlays and five extensions
| Ref | Domain | Statements | Status |
|---|---|---|---|
| D1 | AI foundations | 22 | PUBLISHED |
| D2 | Data and information | 23 | PUBLISHED |
| D3 | Interaction and elicitation | 19 | PUBLISHED |
| D4 | Automation and integration | 22 | PUBLISHED |
| D5 | Solution design | 22 | PUBLISHED |
| D6 | Evaluation and assurance | 22 | PUBLISHED |
| D7 | Ethics, governance and professional practice | 22 | PUBLISHED |
| D8 | Security | 25 | PUBLISHED |
Four proficiency levels are defined: Aware, Applied, Proficient and Advanced. A fifth level is reserved and deliberately undefined.