AI builders and no-code platforms have an accessibility blind spot

AI-generated and drag-and-drop site builders have made shipping a polished-looking website faster than ever. What they haven't solved — and mostly don't attempt to — is whether that polished site actually works for someone using a screen reader, navigating by keyboard, or relying on sufficient colour contrast.

In every audit we've run on a Lovable, Base44, Wix or Webflow site this year, the same handful of issues turn up — and they're the same issues that have failed WCAG audits for a decade, just generated at higher speed now.

What we see most often

• AI-generated heading structures that skip levels or duplicate an h1, breaking screen-reader navigation
• Decorative images with no alt text, or meaningful images with generic auto-generated alt text
• Colour palettes chosen for brand vibe, not contrast ratio — especially light text on brand-colour buttons
• Custom interactive components (drag-and-drop generated) with no keyboard access at all
• Form fields with placeholder-only labels, exactly the pattern we flagged on our own site recently

Why this happens

AI builders optimise for "looks right" and "matches the prompt" — accessibility isn't part of the generation loop unless a platform deliberately builds it in, and few currently do well. No-code visual builders have the opposite problem: they give a lot of layout freedom, and freedom without guardrails tends to produce inaccessible markup by default, because the person dragging elements around usually isn't thinking about the accessibility tree.

Neither failure mode is a reason not to use these tools. It is a reason to treat an accessibility pass as a required step before launch, not an optional extra — especially for public-sector work or any organisation with Equality Act exposure.

What to actually do about it

A proper WCAG 2.2 AA audit catches all of the above and more — including the criteria that automated scanners alone will miss, like whether keyboard focus order actually makes sense. We audit and remediate across every platform we support, AI-built or otherwise, and it's usually a fast, bounded piece of work rather than a rebuild.