We Ran an Accessibility Audit on 154 AI-Built Websites
One text element in five on websites built by Lovable fails WCAG AA contrast. The model chose 70 percent of those colours. We wrote the other 30 percent ourselves.
7
min read

Table of Contents
Two thirds of AI-built sites have a contrast failure. The model picked most of those greys, but a third of them came straight out of the prompt.
We tested 11,734 pieces of text across 154 websites built by Lovable from our Lovable prompts, against the WCAG AA contrast requirement. 2,432 failed, which is 20.7 percent. Two thirds of the sites, 102 of 154, have at least one failure. Only 52 sites are completely clean.
Key findings
One text element in five (20.7 percent of 11,734) fails WCAG AA contrast on websites Lovable built from our prompts.
Two thirds of the sites, 102 of 154, have at least one contrast failure. 52 are completely clean.
The failures are near misses: the median failing ratio is 3.89 against the 4.5 to 1 requirement, and 45 percent sit within 0.5 of the limit.
The model chose 70 percent of the failing colours. The other 30 percent were written in our own prompts.
The structure is sound: every site had one h1 and a lang attribute, and 88 percent of images have alt text.
This is not a story about AI being careless in general. The rest of the audit is better than we expected. It is one specific habit, it is easy to describe, and a meaningful part of it turned out to be our own fault.
How we measured accessibility
We measured between 29 September and 2 October 2026. Each site was built by Lovable from a single prompt, most of them in one build as our Lovable credit study shows, published to a lovable.app address and loaded in Chromium at 1366 by 900. Every paragraph, heading, link, list item, button and label carrying at least four characters of its own text was inspected: we read the computed text colour, walked up the tree for the nearest solid background, calculated the contrast ratio and compared it against the WCAG AA thresholds of 4.5 to 1 for body text and 3 to 1 for large or bold text.
Where text sat on a photograph, a gradient or a semi transparent layer we excluded it rather than guessing, which removed 5,415 elements from the sample. Text over images is a real accessibility problem, but it cannot be judged from the stylesheet alone, and a smaller honest number beats a larger convenient one.
One detail is worth naming because it invalidated our first attempt. These sites define colours in oklch() rather than rgb(). Our first pass parsed those values as if they were RGB and produced an absurd result, with several sites apparently failing every single element. Converting properly brought the figure from 39 percent down to 20.7 percent. If you run this kind of audit yourself, check that before you believe your own output.
What fails most often: near misses on small grey text
45 percent of sampled failures sit within 0.5 of the threshold. Text at 4.15 where 4.5 was required.
The median failing element is 14 px, and just under half of all failures are on text smaller than that.
The median failing ratio is 3.89 against a 4.5 requirement.
The median site has 8 failures, covering 14 percent of its text.

Contrast failures across 154 sites built by Lovable. Most failing text is a near miss rather than a dramatic one, which is why it is so easy to overlook.
Put those together and the picture is clear. This is not white text on a yellow button. It is the caption under the photograph, the date on the blog card, the secondary line in the pricing table, the small print in the footer. It is the text treated as less important, rendered in a grey that looks refined on a designer's monitor and vanishes on a phone in daylight.
The worst offenders are the sites with the most tasteful palettes. A golf club failed 53 of 79 elements. A funeral home failed 52 of 82. An interior design studio failed 37 of 59. Restrained, muted, expensive looking, and hard to read.

A funeral home built from our funeral home prompt. The line under each price renders at 3.99 to 1 on the card colour, below the 4.5 to 1 that body text needs. This is the typical failure: tasteful, small and slightly too light.
Is low contrast the AI's fault or the prompt's?
Our first version assumed the model invented these colours. Because we write the prompts ourselves, we could check, so we did. For every failing piece of text on every site whose prompt defines a palette, we compared the colour the browser painted against the colours the prompt specified.
70 percent of failing text uses a colour the model chose on its own, with no equivalent anywhere in the prompt. We looked at why that happens in what Lovable keeps from your design system.
30 percent uses a colour that was written in the prompt. We specified it, and the model applied it faithfully.
On 24 of the 94 sites with a defined palette and at least one failure, the prompt is the main cause rather than the model.

Where each failing text colour came from, on sites whose prompt defines a palette. Grey is a colour the model picked, orange is a colour we wrote ourselves.
We then audited the palettes themselves rather than the pages. Of the twenty prompts written to our current design system, twelve specify a muted text colour that cannot pass WCAG AA against its own background. The pattern is always the same: the muted grey clears 4.5 against the page colour, then fails against the slightly darker surface used for inset panels, which nobody thought to check.

The muted grey from our nail salon prompt. It clears 4.5 to 1 on the page colour, then fails on the two slightly darker surfaces the same prompt defines. The model applied it exactly as written, so this one is on us.
So the model is responsible for most of this, but not for all of it, and the part we own is the part we can fix with certainty. A colour you wrote down is a colour you can test before you press send.
What AI-built websites get right
Every single site had exactly one h1 and a lang attribute on the document. Not one of the 154 got this wrong.
88 percent of images carry alt text, and only 13 sites had an image missing one.
86 percent of images use native lazy loading without being asked, which matters because images are most of what these sites weigh.
Only 10 percent of sites overflow horizontally at 375 px width.

The structural checks almost all pass. Tap targets are the exception: only 4 of 154 sites keep every link and button at 24 px or larger.
The one structural weak spot is touch targets. 150 of 154 sites have at least one link or button under 24 px in one dimension, with a median of 11 per site. Most are text links in a navigation bar or footer, which is the same pattern you find on hand-built sites, but it still deserves a line in the prompt.
How to make an AI-built website pass WCAG AA
Because the cause is narrow, so is the fix, and all four lines belong in the prompt itself. If you are starting from scratch, our guide on how to write an AI website prompt covers the rest of the structure. The first line is the one we had to apply to our own library.
Check the colour you are about to specify. Your muted text colour has to clear 4.5 to 1 against every surface it will sit on, not just the main page colour. This is arithmetic, not taste, and it catches the third of the problem that starts with you.
Give the muted colour a floor in the prompt too. State it as a requirement: every text colour, including secondary and caption text, must reach at least 4.5 to 1 against the surface it sits on. Name the muted token explicitly, because that is the one that fails.
Set a minimum text size. Nothing below 14 px, and captions at 13 px only where they also clear 4.5 to 1. Half the failures live under that line.
Set a minimum tap target. Interactive elements at least 24 by 24 px, or 44 by 44 px to meet the stricter mobile guidance.
There is a slightly embarrassing proof that the model will comply when asked. One project in our library needed a second AI pass, and its commit message reads, in full, Contrasted hero text on dark. It fixed the contrast the moment it was told to. It simply did not think to do it first, and neither did we.
Why contrast matters beyond compliance
Accessibility arguments usually get made on legal grounds, which is fair and completely unpersuasive to someone building a page for their own small business. So here is the practical version.
A contrast ratio of 4.15 is not a technicality. It is the difference between a price being read and being skipped, on a phone, at arm's length, in a bright room, which is where a large share of your visitors are. The audit above is really a readability audit that happens to have a standard attached.
And unlike most of what a prompt controls, this one is free. Lifting a muted grey from 4.1 to 4.6 costs nothing, changes nothing anyone will notice in a side by side, and makes the page work for the visitor standing in the sun.

A golf club built from our golf club prompt, as built and with the same grey darkened just enough to pass. Side by side the change is hard to spot, but the text moves from 4.21 to 4.65 to 1.
The newer prompts in our library of website prompts state this contrast floor explicitly, so you can start from one that already does. In our own library the corrections were shifts of at most twelve points of brightness, with the hue untouched.
Update, 2 October 2026. An earlier version of this article blamed the model for every failing grey. We then traced each colour back to the prompt, and a third of them were ours. The numbers above include that correction.

