Somewhere on the internet right now, a blind person is trying to check a payroll balance, and a small floating icon — a stick figure in a circle, usually — is making that harder. The icon is an accessibility widget. It was installed by the employer, or the payroll vendor, to make the site accessible. The employee already has a screen reader, configured over years to their exact preferences: speech rate, verbosity, keyboard shortcuts. The widget does not know this. It rewrites the page on the fly, re-labels things, hijacks focus, and occasionally picks a fight with the screen reader it was ostensibly installed to help. The employee files a complaint. The company is baffled. They bought accessibility. It renews monthly.
This is the strange economics of the accessibility overlay, a product category built on a single seductive pitch: paste one line of JavaScript into your site and become compliant with disability law. No redesign, no remediation, no awkward conversations with your web team. Just a script tag and a subscription. The pitch has attracted thousands of business customers and an unusually organized opposition — because overlays are a rare case in software where the person who pays for the product and the person who has to live with it are different people, and the product only has to satisfy one of them.
Follow the incentive and the whole category snaps into focus. Web accessibility in the United States is enforced largely through litigation: thousands of lawsuits a year, many of them template-driven, targeting businesses whose sites trip up assistive technology. For a small business owner, the demand letter arrives like weather — arbitrary, expensive, vaguely menacing. Into that anxiety steps the overlay vendor, selling not accessibility exactly but litigation weatherproofing. Compared to a proper remediation — auditing templates, retraining content editors, rebuilding a checkout flow — a widget subscription is pocket change. The buyer is not purchasing a better experience for disabled visitors. The buyer is purchasing an answer to the question “what are you doing about accessibility?” and the widget is a very good answer to that question, in the same way a certificate framed in the lobby is a good answer.
The Federal Trade Commission eventually said the quiet part in writing. In its 2025 complaint against accessiBe, the biggest name in the overlay business, the agency described the company’s marketing of an “AI-powered” solution that, with one line of code, would handle 30 percent of WCAG requirements instantly and the remaining 70 percent within 48 hours. The complaint alleged that sites running the widget still failed basic WCAG criteria across industries, and cited the W3C and WebAIM for a position that gutted the whole premise: no automated tool can determine conformance on its own. Manual testing is required. Which is to say, the thing the overlay promises to replace is the thing that cannot be replaced.
The people who actually use assistive technology figured this out early, and they said so with unusual coordination. The Overlay Factsheet, a community statement signed by hundreds of accessibility specialists — more than 700 by 2022, including people employed by Google, Apple and Mozilla — concludes flatly that full compliance cannot be achieved with an overlay, that automated repair of text alternatives, form labels and keyboard access is unreliable, and that overlays do nothing for PDFs, canvas, SVG or media. Then it makes the subtler point: the widget is at best redundant, because disabled users already carry their tools with them. A blind person’s screen reader lives on their device, tuned to their needs. The overlay is a second, worse screen reader imposed by the site, sometimes overriding the first. It is accessibility as imagined by someone who has never needed it: a feature you install on a building rather than a capability people bring through the door.
And the weatherproofing leaks. The Seattle Times reported in 2022 that more than 400 companies running an accessibility widget or overlay were sued the previous year anyway. One community-maintained appendix lists over 200 companies sued in 2021 despite their overlays. The product’s core promise — make the lawsuits go away — fails on its own terms, which would normally kill a category. It has not, because the customer’s alternative is real work, and the lawsuit risk is probabilistic, and hope renews monthly at a price below the threshold of scrutiny.
The case that best captures the inversion is the one the Seattle Times reported between LightHouse for the Blind and Visually Impaired and the payroll giant ADP: blind employees locked out of workplace systems after an overlay conflicted with their screen readers. The settlement reportedly included ADP agreeing not to rely solely on overlays. Read that again. The accessibility tool became the accessibility barrier, and the resolution was a promise to stop depending on the tool. The American Foundation for the Blind now warns that overlays can override users’ own assistive settings and miss the majority of WCAG guidelines that require human judgment.
None of this means the underlying problem is small. WebAIM’s annual scan of the top million home pages finds detected WCAG failures on roughly 96 percent of them — and automated scans catch only a subset of failures, so the true figure is worse. The web is genuinely, structurally hostile to disabled users, and the businesses responsible for it are genuinely, structurally uninterested in the slow fix. The overlay is what a market produces when legal risk is real, expertise is scarce, and the people harmed are not the people paying. It is compliance theater with a settings panel.
The tell is in the icon itself. A site that was actually built accessibly needs no badge announcing it; the accessibility is invisible, baked into the markup, indistinguishable from competence. The little stick figure in the corner exists precisely because the work wasn’t done. It is not a sign that a site is accessible. It is a sign that someone, at some point, was asked what they were doing about accessibility — and found the cheapest possible answer.