Control · C-079 · Fairness & Customer Outcomes
Accessibility and inclusive design review
AI-driven channels are designed and tested with assistive technology and with people who have access needs before release, and an equivalent alternative route is kept available.
- Control type
- Preventive
- Risks addressed
- 2
- Audience
- Heads-of · Practitioner
- Frameworks
- Section 20 · PRIN 2A.6 · GOVERN 3
- Responsible AI domain
- Lifecycle stage
- Reviewed
- 2026-11-11
What it does
It checks that the people an organisation serves can use what it has built. Before release, the channel is tested against accessibility standards, with the assistive technologies customers use, and by people with a range of impairments and levels of confidence with technology. Barriers are removed or an equivalent alternative is provided. The business owner of the channel is accountable; design and engineering carry out the work. It operates at design, before each release and after significant change.
How to implement it
- Identify who uses the channel and what access needs they have.
- Test with screen readers, voice control, magnification and keyboard-only use.
- Test speech and image features with people whose voice, appearance or movement is atypical.
- Allow more time, simpler language and other adjustments on request.
- Keep a human or other alternative that gives the same result, and make it easy to find.
- Record known barriers and the plan for each.
Evidence it operates
Design: accessibility requirements for AI channels. Operation: test reports including sessions with disabled users; a log of barriers and fixes; evidence that the alternative route is available and used; adjustments recorded for individual customers.
Where it fits
At design and before release. It is the primary control for exclusion and inaccessibility and supports unequal performance across groups where the cause is an impairment. It forms part of pre-deployment testing and acceptance.
Limits
Standards written for web pages do not fully cover conversational and voice interfaces. Testing with a few people cannot represent the range of impairments. The underlying model's weaknesses, such as poorer recognition of atypical speech, may not be fixable by the organisation, which makes the alternative route essential and not optional.
Risks this mitigates
Primary mappings
- Exclusion and inaccessibility Medium
- Tests AI channels with assistive technology and affected users and keeps an alternative. An AI-driven channel cannot be used effectively by disabled people, or by others the organisation has a duty to serve, and no equivalent alternative is available.
Supporting mappings
- No route to a human Medium
- Confirms the route works for people the channel serves poorly. A customer cannot reach a person when the automated channel fails them, or the route to one is hidden or obstructed.
Framework and clause references
| Framework | Clause | Title |
|---|---|---|
| Equality Act 2010 | Section 20 | Duty to make adjustments Duty to make adjustments. |
| FCA Consumer Duty (PRIN 2A) | PRIN 2A.6 | Consumer support outcome No unreasonable barriers. |
| NIST AI Risk Management Framework (AI RMF 1.0) | GOVERN 3 | Workforce diversity, equity, inclusion and accessibility in AI risk decisions Accessibility. |
Related terms
How to cite
Axiom Verity (2026). Accessibility and inclusive design review. Reference Library, C-079. https://axiom-verity.com/controls/accessibility-and-inclusive-design-review. Reviewed 2026-11-11.
Found an error or a better source? Report a correction.