Risk · R-115 · Fairness, Bias & Impact
Exclusion and inaccessibility
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.
- Severity
- Medium
- Audience
- Heads-of · Practitioner
- Frameworks
- Section 20 · Section 29 · PRIN 2A.6 · GOVERN 3
- Responsible AI domain
- Lifecycle stage
- Reviewed
- 2026-11-11
What it is
Replacing a human channel with an automated one changes who can use it. A chat assistant may not work with a screen reader. A voice system may not understand someone with a speech impairment. An identity check may defeat a person with a tremor or a facial difference. A conversational interface may be unusable for someone with a cognitive impairment or limited English. Organisations that serve the public have a duty under the Equality Act 2010 to make reasonable adjustments for disabled people, and the duty is anticipatory. The risk is that the AI channel becomes the only practical route and some people cannot take it.
How it happens
- The channel was designed and tested with no disabled users.
- The human alternative was withdrawn or made hard to find.
- Accessibility was checked for the website and not for the assistant.
- Time limits and identity steps assume a typical user.
Why it matters
People are shut out of something they are entitled to. For the organisation it is a breach of a legal duty, and for regulated firms it conflicts with the expectation to support customers, including those with characteristics of vulnerability.
Warning signs
- No accessibility testing of the AI channel is on record.
- Abandonment is high among customers with registered access needs.
- The telephone or branch alternative has been reduced.
- Complaints mention being unable to get past the automated system.
Mitigating controls
Primary controls
- Accessibility and inclusive design review Preventive
- Tests AI channels with assistive technology and affected users and keeps an alternative. 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.
Supporting controls
- Customer outcome monitoring Detective
- Shows whether some groups abandon or fail the journey more often. The outcomes customers receive from AI-driven journeys are measured, broken down by customer group and compared with other channels, and poor outcomes are investigated and acted on.
- Escalation to a human Preventive
- Gives people the channel does not serve a route to a person. Automated channels offer a prompt, unobstructed route to a person, available on request and triggered automatically by defined signals, with the conversation handed over.
Framework and clause references
| Framework | Clause | Title |
|---|---|---|
| Equality Act 2010 | Section 20 | Duty to make adjustments The duty to make adjustments. |
| Equality Act 2010 | Section 29 | Provision of services, etc. Duties of those who provide a service to the public. |
| FCA Consumer Duty (PRIN 2A) | PRIN 2A.6 | Consumer support outcome Support without unreasonable barriers. |
| NIST AI Risk Management Framework (AI RMF 1.0) | GOVERN 3 | Workforce diversity, equity, inclusion and accessibility in AI risk decisions Accessibility in AI risk decisions. |
Sources
- NIST AI Risk Management Framework (AI RMF 1.0) NIST · framework
How to cite
Axiom Verity (2026). Exclusion and inaccessibility. Reference Library, R-115. https://axiom-verity.com/risks/exclusion-and-inaccessibility. Reviewed 2026-11-11.
Found an error or a better source? Report a correction.