WCAG 3.0 is an active W3C Working Draft in development by the AGWG. Requirements, assertions, the conformance model and reporting tiers are still being developed and may change. The latest published draft is the 10 September 2026 Working Draft; this site follows the editors’ draft, which also includes exploratory provisions.
An organisational assertion: the organisation states that its authors train AI models on representative, unbiased disability-related data, in proportion to the general population.
Why It Matters
Models trained on data where disabled people are rare or stereotyped reproduce that in their output - and increasingly that output is the interface.
0
Test Procedure from the W3C Draft
Upstream has a test heading for this provision but has not written
the procedure yet, and no test steps have been written here.
Generic advice would apply to almost any provision, so none is shown.
W3C Working Group Status
W3C Maturity Status:developing
Track / Group:policy-and-protection
The draft records no open editors' notes on this provision.
W3C Threads on This Provision
No GitHub issue in w3c/wcag3 carries this provision's label, so there
is no recorded discussion to show. Cross-cutting threads that name no
provision are listed on the
Discussions page.
WCAG 2.2 Criteria That Map Here
From this site's reviewed WCAG 2.2 → WCAG 3 map. The map is
an interpretation, not a W3C mapping: W3C has not published one yet.
No mapping recorded. No WCAG 2.2 criterion in
this site's map is recorded as a predecessor of this provision.
[Title, role, or organization] asserts that:
* Content author(s) train AI models using representative and unbiased disability-related information that is proportional to the general population.