An ADA website lawsuit is a claim that a business shut disabled people out of its website or app, most often because a screen reader cannot make sense of it. In 2025 plaintiffs filed 3,117 of these suits in US federal court alone, most against small companies, and AI coding tools reproduce exactly the errors those suits cite. Best Answer Hub wrote this playbook for founders who built something real with AI tools: what the risk actually is, which parts of the viral warnings hold up, and how to find and fix the problems without reading a line of code.
This playbook is the accessibility half of a pair. Its companion, Can Your Vibe-Coded App Get You Sued?, covers privacy, tracking, marketing messages, children's data, and subscriptions. Both feed one free instrument: the Best Answer Hub Vibe Coder Readiness Check, whose compliance dimension scores this exact ground.
This is plain-English education, not legal advice. Best Answer Hub is not a law firm, and the law here varies by court and by state. If you have already received a demand letter or a lawsuit, talk to a lawyer before you reply, pay, or change anything on the site, because what you say and do next can matter to the outcome.
Does the ADA apply to my website or app?
For any business open to the public, the US Department of Justice says yes. Its March 2022 guidance states that the Americans with Disabilities Act applies to the web offerings of businesses covered by Title III, and that businesses "have flexibility in how they comply." There is still no formal federal regulation setting a technical standard for private websites, so the guidance points to the Web Content Accessibility Guidelines (WCAG) as helpful, not mandatory. That gap is why the courts disagree with each other.
The disagreement is about websites with no physical location. The Ninth Circuit held in Robles v. Domino's (2019) that the ADA covered Domino's website and app because they connected to its restaurants, and the Supreme Court declined to review it. A California appeals court held in 2022 that a website-only business is not a place of public accommodation. The Eleventh Circuit's 2021 ruling against coverage in Gil v. Winn-Dixie was later vacated as moot, which left Florida without a settled answer. For a founder the practical reading is simple: lawsuits are filed in every one of these places, defending even a weak one costs money, and building accessibly costs far less. Best Answer Hub plans on the assumption that coverage applies.
How many ADA website lawsuits are filed, and where?
About 3,000 a year in federal court, with a sharp rebound in 2025. Seyfarth Shaw, a law firm that has tracked ADA Title III filings for years, counted 3,255 federal website accessibility suits in 2022, 2,794 in 2023, 2,452 in 2024, and 3,117 in 2025, a 27 percent rise that it attributes partly to new plaintiffs' firms entering the field. Website suits made up 36 percent of all 8,667 federal Title III suits in 2025.
Source: Seyfarth Shaw, federal court website accessibility lawsuit filings, March 2026. Federal courts only; state-court suits and demand letters are not counted.
Federal counts are a floor, not the full picture. UsableNet, an accessibility vendor that also tracks New York and California state courts, counted more than 5,000 digital accessibility lawsuits in 2025 across federal and state courts. Its report adds three numbers that matter to a small founder: 64 percent of suits named companies with under $25 million in revenue, 45 percent named companies that had already been sued before, and e-commerce sites drew about 70 percent of filings. Demand letters, the step before a suit, are not counted by anyone, and Seyfarth notes that a significant number never reach a court at all.
Geography is concentrated. In 2025 federal website suits, New York led with 1,021, Florida followed with 961, and Illinois with 585. California logged just 4, which looks like safety and is not: plaintiffs there file in state court under the Unruh Civil Rights Act, which carries its own damages. Activity has kept climbing in 2026, and UsableNet projects roughly 6,000 digital suits for the full year, about a fifth more than 2025. Best Answer Hub reads all of this as one message: the volume is real, and it lands mostly on small sites.
Is the viral "$4,000 letter from New York" claim true?
Two of its three claims hold up, and the money detail is borrowed from another state. A short video circulating among vibe coders says a demand letter "from New York" asks for $4,000, that over 3,000 such cases reached federal court last year, and that New York and Florida filed the most. The table below checks each claim against the primary data.
| The viral claim | Verdict | What the record shows |
|---|---|---|
| "Over 3,000 went to federal court last year" | True | 3,117 federal website suits in 2025 (Seyfarth Shaw); 3,195 federal digital suits (UsableNet). |
| "New York and Florida filed the most" | True for 2025 | New York 1,021 and Florida 961 federal website suits in 2025. Counting all Title III suits in early 2026, California leads. |
| "A letter from New York asking for $4,000" | Misattributed | $4,000 is California's Unruh Act minimum per offense. No New York statute sets that figure. |
| "A picture with no label is enough for a letter" | Plausible | Missing alt text appears on 53.1% of top home pages (WebAIM, 2026) and is a standard complaint item. |
| "Test it with a screen reader" | Good advice | Automated tools miss a large share of barriers; a free screen reader catches what they cannot. |
The $4,000 figure comes from California Civil Code section 52, which sets damages "in no case less than four thousand dollars" for each offense under the Unruh Act, and an ADA violation counts as an Unruh violation. Plaintiffs once argued that meant $4,000 per visit; in the Domino's case a federal court treated the inaccessible website as one overarching violation and limited the award to $4,000. New York state and city human rights laws allow damages too, but they set no $4,000 floor. Typical settlement amounts are harder to pin down: the published ranges, often quoted between $5,000 and $20,000 for a website case, come from vendors who sell compliance services, and Best Answer Hub found no independent data behind them.
Do AI coding tools write accessible code?
Not by default, even when asked to. AudioEye, an accessibility vendor, asked five AI tools (from OpenAI, Anthropic, Google, xAI, and Lovable) to build three sites each, a news site, a store, and a financial services site, all to WCAG 2.2 level AA. All 15 sites failed basic Level A checks. The study logged 306 distinct issues occurring more than 59,000 times, an average of 55 issues per AI-built page, with a third of issues blocking a task outright and nearly half visible only when the site is actually used. It is vendor research, published in 2026, and Best Answer Hub labels it that way, but its direction matches independent work.
That independent work points the same way. The WebAIM Million, an annual scan of the top one million home pages, found detectable WCAG failures on 95.9 percent of them in 2026, up from 94.8 percent, with average errors per page rising 10 percent to 56.1. WebAIM links the rise to heavier frameworks and AI-assisted "vibe coding," which is its interpretation rather than a direct measurement. A 2026 academic study of 300 interfaces generated by Claude, Gemini, and GPT models found about 1.8 semantic accessibility problems per interface that automated checkers pass, such as images labeled simply "image."
Language models are not uniquely bad at this. A 2025 comparative study found they often produce more accessible code than human developers on basics such as color contrast and alt text, while struggling with complex patterns like ARIA attributes. The lesson is not "AI ruins accessibility." It is that a prompt asking for accessibility is a request, not a result, and that the person who ships the app owns the outcome either way.
What do ADA demand letters actually target?
Mostly the same six errors, because they are the ones automated scans find on almost every site. The WebAIM Million 2026 reports that 96 percent of all detected errors fall into six categories, and they map closely to the barriers complaints describe: a screen reader user who cannot tell what a button does, cannot fill in a form, or cannot read a low-contrast price.
Source: WebAIM Million, 2026 edition (data from February 2026). These six categories account for 96 percent of all detected errors.
Each one has a plain-English fix and a one-line instruction for a coding agent. Best Answer Hub recommends handing the agent all six as a standing rule rather than one at a time, then asking for proof.
- 1Low-contrast text (83.9 percent). Gray text on a white background, or white text on a pale button, that low-vision users cannot read. WCAG AA asks for a 4.5 to 1 contrast ratio for normal text. Tell the agent: "Make every text and background pair meet WCAG AA contrast." The free Best Answer Hub Color Palette Generator checks contrast before you pick a color.
- 2Missing alt text (53.1 percent). Images a screen reader announces as nothing, or as a file name. Tell the agent: "Give every meaningful image alt text that says what it shows, and mark decorative images with empty alt."
- 3Missing form labels (51 percent). Inputs that rely on placeholder text, which vanishes as the user types and is often not announced. Tell the agent: "Every input needs a visible label linked to it."
- 4Empty links (46.3 percent). Icon-only links, such as a logo or a social icon, with no text a screen reader can say. Tell the agent: "Every link needs an accessible name."
- 5Empty buttons (30.6 percent). The close X, the hamburger menu, the heart icon. Tell the agent: "Every icon button needs an aria-label that says what it does."
- 6Missing page language (13.5 percent). Without it, a screen reader may read English with the wrong pronunciation rules. Tell the agent: "Set the lang attribute on the html element."
Two newer WCAG 2.2 rules catch apps in particular. Touch targets should be at least 24 by 24 pixels, so tiny icon buttons fail, and a focused element must not be completely hidden behind a sticky header or a cookie banner. Both are easy for an agent to fix once named, and both are invisible to a founder who only ever uses a mouse.
How do I test accessibility if I cannot read code?
Use three free tests that together cover far more than any single tool. Automated scanners are fast but partial: Deque, which makes the axe engine, reports that its automated testing finds about 57 percent of issues by volume, and a UK Government Digital Service experiment that planted 143 barriers on one page found the best single tool caught only 37 to 41 percent, with 29 percent missed by every tool tested. A keyboard and a screen reader fill most of the gap.
- 1The keyboard-only test. Put the mouse away. Press Tab to move through the page and Enter or Space to activate things. Every link, button, menu, and form field should be reachable, in a sensible order, with a visible outline showing where you are. If you get stuck, lose track of focus, or cannot close a pop-up, keyboard and switch users hit the same wall.
- 2The screen reader test. Turn on VoiceOver (built into every Mac and iPhone) or install NVDA, which is free and open source for Windows. Close your eyes and try to sign up, find the price, and complete the main task. Listen for buttons announced as just "button," images announced as file names, and fields with no name.
- 3The automated scan. Run the free axe DevTools or WAVE browser extension, or the Lighthouse accessibility audit built into Chrome, on every key screen, not just the home page. Fix everything it flags, then repeat tests 1 and 2, because a clean scan is where testing starts, not where it ends.
| Test | Cost | Catches | Misses |
|---|---|---|---|
| Automated scan (axe, WAVE, Lighthouse) | Free | Contrast, missing alt, empty buttons and links, missing labels | Meaning: alt text that exists but says nothing useful |
| Keyboard only | Free | Focus traps, unreachable menus, invisible focus, broken pop-ups | What a screen reader announces |
| Screen reader (VoiceOver, NVDA) | Free | Confusing names, reading order, unlabeled controls in context | Visual issues such as contrast |
| Professional audit | Paid | Full WCAG review with real assistive technology users | Little, if done well |
For apps that hold payments, health, or other sensitive tasks, or that serve a large public audience, Best Answer Hub suggests budgeting for a professional audit once the three free tests pass. A Lighthouse score of 100 is a useful signal, but it scores only automated audits, so it can sit on top of an app a blind user still cannot finish.
An agent saying "accessible" is a claim. A clean scan plus a keyboard run plus a screen reader run is evidence.
Will an accessibility overlay widget protect my app?
No reliable evidence says it will, and a federal regulator has penalized the biggest name selling that promise. Overlays are scripts that add a toolbar or automated fixes on top of a site. In April 2025 the US Federal Trade Commission approved a final order requiring accessiBe to pay $1 million, alleging it falsely claimed its widget could make any website compliant with WCAG and formatted third-party reviews to look independent. UsableNet's 2025 data shows lawsuits filed every month against companies already using such widgets.
The accessibility community is blunt about it: the Overlay Fact Sheet, signed by more than 1,000 accessibility practitioners, advocates removing overlays altogether, largely because screen reader users already have their own tools and the overlay does not change the underlying code they read. Best Answer Hub recommends spending the widget's subscription money on fixing the six errors above instead.
Does this apply if my business is outside the US?
US lawsuits follow sales more than headquarters. A 2017 New York federal suit named an Italian company over its website, though that company also ran a New York showroom. New York federal courts have since held that a website merely reachable from the state is not enough for a lawsuit there, while actual sales or deliveries to New York residents can be. A founder anywhere who sells to US customers should treat the risk as real and ask a lawyer about specific exposure.
Europe has its own rule. The European Accessibility Act has applied since 28 June 2025 to e-commerce and other consumer services sold in the EU, and in-scope services must publish information on how they meet the accessibility requirements. Microenterprises providing services, meaning fewer than 10 staff and no more than EUR 2 million in annual turnover or balance sheet, are exempt from those requirements, which covers most solo founders. Products are not exempt, and growth can end the exemption, so Best Answer Hub suggests building to WCAG 2.2 AA now rather than retrofitting later.
Accessibility is one of several ways a vibe-coded app draws legal attention without ever being hacked. Tracking scripts, session recording, marketing texts and emails, children's sign-ups, and subscription checkouts each carry their own rules and penalties. The companion Best Answer Hub playbook, Can Your Vibe-Coded App Get You Sued?, fact-checks those claims with the real figures.
What should I do about accessibility this week?
Five actions close most of the common gaps in an afternoon, none require reading code, and each ends with something you can see or hear yourself. They map to the errors that appear in demand letters most often, and to the compliance dimension of the readiness check below.
- 1Give your agent the six-rule standing order. Paste the six instructions from the checklist above into your agent's project rules file, such as CLAUDE.md or AGENTS.md, so they apply to every future screen, not just today's fix.
- 2Scan every key screen. Sign-up, sign-in, pricing, checkout, and the main task. Ask the agent to fix every axe violation and show you the clean rerun. Passing looks like zero violations on each screen, with the scan saved.
- 3Do one keyboard-only run. Complete sign-up and the main task without touching the mouse. Passing looks like reaching every control with a visible focus outline and no traps.
- 4Do one screen reader run. With VoiceOver or NVDA, complete the same task with your eyes closed. Passing looks like every button and field announcing a clear name.
- 5Remove any overlay widget and add a contact route. Replace the widget with real fixes, and publish an email address where users can report a barrier. A user who can reach you is a user who may not need a lawyer.
Those five turn "my agent says it is accessible" into evidence you produced yourself. The next step is seeing where accessibility sits among everything else an app needs before strangers rely on it, which is what the free score below measures.
Take the free Vibe Coder Readiness Check
45 questions across nine dimensions, including compliance, accessibility, and privacy, with a 0 to 100 score in about 10 to 12 minutes. No signup and no email, and the score is never paywalled.
Score your app nowCommon questions about ADA website lawsuits
Sources
- US Department of Justice, Guidance on Web Accessibility and the ADA, March 2022 (ADA applies to web offerings of public accommodations; flexibility in how to comply).
- Seyfarth Shaw (Launey and Vu), Federal Court Website Accessibility Lawsuit Filings Bounce Back in 2025, March 2026 (3,255 / 2,794 / 2,452 / 3,117 federal website suits, 2022 to 2025; 2025 state counts).
- UsableNet, 2025 Year-End Digital Accessibility Lawsuit Report (more than 5,000 federal and state suits; 64 percent under $25M revenue; 45 percent repeat defendants; about 70 percent e-commerce; vendor report).
- UsableNet, 2025 Midyear Accessibility Lawsuit Report (suits filed every month against sites using accessibility widgets).
- Equidox, 2026 midyear accessibility lawsuits (UsableNet projection of more than 6,000 digital suits in 2026).
- California Civil Code section 52, Unruh Civil Rights Act damages ("in no case less than four thousand dollars").
- Bryan Cave Leighton Paisner, Domino's website violates the ADA; Unruh penalties limited to $4,000, June 2021.
- Hunton Andrews Kurth, California Court of Appeal narrows reach of ADA and Unruh for e-commerce, 2022 (Martinez v. Cot'n Wash).
- SCOTUSblog, Domino's Pizza v. Robles (certiorari denied October 2019).
- Justia, Gil v. Winn-Dixie, vacated as moot, 11th Cir., December 2021.
- Classaction.org, Norman v. Minotti S.p.A. complaint, S.D.N.Y., 2017 (Italian company sued over its website).
- WebAIM, The WebAIM Million, 2026 edition (95.9 percent of home pages with detected failures; 56.1 errors per page; six most common failures).
- AudioEye, AI coding tools do not write accessible code, 2026 (15 AI-built sites, all failed Level A; 55 issues per page; vendor research).
- Calò, Guriţă and De Russis, Measuring the Semantic Accessibility Gap in LLM-Generated Web UIs, CHI EA 2026 (about 1.8 semantic violations per generated interface).
- Human or LLM? A Comparative Study on Accessible Code Generation Capability, arXiv:2503.15885, 2025.
- W3C, What's New in WCAG 2.2 (target size minimum, focus not obscured, accessible authentication).
- Deque, Automated testing study identifies 57 percent of digital accessibility issues, 2021 (vendor research).
- UK Government Digital Service, Testing tools on the world's least-accessible webpage, 2017.
- Chrome for Developers, Lighthouse accessibility scoring (automated audits only; manual audits do not count).
- NV Access, About NVDA (free, open-source screen reader for Windows).
- US Federal Trade Commission, FTC approves final order requiring accessiBe to pay $1 million, April 2025.
- Overlay Fact Sheet (more than 1,000 practitioner signatories).
- Directive (EU) 2019/882, European Accessibility Act, Article 4(5) microenterprise exemption and Article 3 definitions.
More from Best Answer Hub: Can Your Vibe-Coded App Get You Sued?, Is Your Vibe-Coded App Ready to Launch?, the Color Palette Generator, and the Knowledge Center assessments.