As of 13 August 2026, AI can only partly check whether your website meets WCAG requirements.
This still needs a person who signs their name to it.
Can you do it?
15 minutesto a draft.
2 hoursto something you’d act on.
Cost, all in£0
Skill neededpower-user
Who has to check ita professional
What the alternative costsA specialist accessibility audit is the alternative, but no price is provided in the supplied data.
If this goes wrong: you publish a site as accessible, leave barriers in place and face remediation work, lost users or a complaint.
What to actually do
Hand it to a person
The route this page recommends
Someone with a licence or accountable authority has to sign this before it counts.
Use a tool built for this
Second choiceDo it yourself
The distant thirdA chat interface, power-user skill, and roughly 2 hours until you can act on the result.
How to actually do it
- Write down the WCAG version, target conformance level, pages, forms and user journeys that the check must cover.
- Open each representative page and record its URL, visible content, interactive states, authentication requirements and any important mobile or responsive views.
- Run an automated accessibility checker on the in-scope pages, then save its findings, screenshots and test settings without treating warnings as proof of compliance.
- Gather the relevant HTML, CSS, JavaScript, component code, design content and error messages, removing secrets and personal data before pasting them into the prompt.
- Paste the scope, checker results and gathered material into the prompt and ask the AI to produce the evidence-based findings report.
- Test the reported journeys with only a keyboard, checking focus order, focus visibility, traps, labels, errors and status messages, then mark each result against the AI report.
- Send unresolved findings and the retest plan to an accessibility specialist, and update the site only after comparing each fix against the original reproduction steps.
Prompt
Assess the accessibility of the website materials below against [WCAG version] at [target conformance level] for [scope of pages, user journeys and content]. Do not claim that the website complies based only on automated checks. Separate the report into: 1) confirmed failures supported by the supplied evidence, 2) likely issues needing manual testing, 3) items that passed the supplied checks, and 4) areas you could not assess. For every finding, give the relevant success criterion, affected page or component, reproduction steps, who may be affected, severity with a short reason, and a practical remediation. Distinguish missing evidence from an actual failure, flag any ambiguous interpretation, and do not invent page behaviour, test results or legal conclusions. Finish with a prioritised retest plan and a list of checks that require keyboard use, zoom, focus testing, screen-reader testing or review by an accessibility specialist. Website URL: [URL] Pages and user journeys in scope: [SCOPE] Target WCAG version and conformance level: [TARGET] Automated checker results: [PASTE RESULTS OR WRITE NONE] HTML, CSS, JavaScript or component code: [PASTE RELEVANT MATERIAL] Screenshots, page text and interaction notes: [PASTE MATERIAL] Known authentication, payment, form or dynamic states: [DESCRIBE THEM OR WRITE NONE]
Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.
What it gets wrong
- AI cannot prove that every page, state, user journey and content change has been tested.
- Automated results miss barriers involving meaning, interaction flow, timing, wording and some assistive technology behaviour.
- AI cannot reliably reproduce the experience of disabled users across browsers, devices and assistive technologies from a code sample alone.
- A polished report does not transfer responsibility for an accessibility statement, a compliance claim or a complaint.
- AI cannot decide every ambiguous success-criterion interpretation without the product context and an informed accessibility judgement.
What caps this at PARTLY: verification cost, judgement under ambiguity and legal accountability.
How we scored this
Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.
| Axis | Score (0–2) |
|---|---|
| Output | 2 |
| Inputs | 1 |
| Verification | 1 |
| Liability | 1 |
| Effort delta | 1 |
| Total | 6 / 10 |
The methodology and its thresholds are published in full.
FAQ
- Can ChatGPT check if my website is WCAG compliant?
- It can review supplied pages, code, screenshots and automated test results, then organise likely failures and suggested fixes. It cannot prove full WCAG conformance, so a compliance claim needs manual testing and, for a serious release, an accessibility specialist.
- Can AI find accessibility issues on my website?
- Yes, AI can identify many issues in markup, content, focus handling and automated checker results. It will miss some problems and can misread incomplete evidence, so test the affected journeys yourself and have unresolved findings checked by a specialist.
- Is an automated accessibility scan enough for WCAG compliance?
- No. An automated scan is useful for repeatable checks, but it does not assess every content, interaction, keyboard, zoom or assistive technology requirement.
- Do I need an accessibility expert to check my website?
- For an initial review, you can use AI alongside automated checks and basic keyboard testing. For a defensible conformance assessment or a high-stakes public service, an accessibility expert should verify the scope, manual tests, evidence and final conclusions.
Nearby answers
Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.
The newsletter
AI news, new answers and product picks, straight to your inbox.