Skip to content
wcagc
←Back to blog
ADATitle IIIlawsuits

AI accessibility lawsuits: what the 2026 trend means

AI-assisted complaints are driving more ADA accessibility lawsuits. What the 2025 pro se data shows, and what actually lowers your risk.

P

Pavel Charkasau

AI didn't invent ADA website lawsuits, but it has lowered the cost of filing one. The clearest signal is in the self-represented numbers: in the first nine months of 2025, pro se plaintiffs filed 1,867 federal ADA Title III cases, already more than the 1,774 they filed in all of 2024, which works out to a 40% rise on a monthly-average basis (Seyfarth ADA Title III). Seyfarth credits AI-assisted complaint drafting for much of that jump. A person who once needed a lawyer to write a complaint can now generate a passable one in an afternoon, which widens the pool of people who can file. That is part of why the overall count keeps climbing: 3,117 federal website accessibility suits in 2025, up 27% from the year before (Seyfarth). For your site the practical meaning is narrow. AI changes who files and how fast, not what they allege. Template complaints still name the same machine-detectable barriers, on the same pages. Below is what the data shows, how these complaints are drafted, and what lowers your risk under the ADA.

How are AI tools changing ADA website lawsuits?

They lower the cost of bringing a case. Filing an ADA Title III complaint used to mean paying a firm or finding one willing to work on contingency. General-purpose AI tools like ChatGPT, Copilot, and Gemini change that math: someone who encounters an inaccessible checkout can now ask a chatbot to draft a federal complaint and file it themselves (Seyfarth ADA Title III).

That is a supply-side shift, not a change in the law. The Americans with Disabilities Act reads the same as it did in 2024. What changed is how many people can produce a filing that clears the bar to get docketed. When the cost of an action drops, the volume goes up. That is the whole story behind the pro se surge.

How much have pro se filings actually risen?

By 40% for federal ADA Title III, on a monthly-average basis. Seyfarth's October 2025 analysis counted 1,867 pro se ADA Title III cases in the first nine months of 2025, against 1,774 for the entirety of 2024 (Seyfarth ADA Title III). Nine months already outran twelve. The parallel Fair Housing Act figure moved further: pro se FHA filings rose 69% over the same comparison.

Those pro se cases sit inside a larger website-accessibility total that is also up. Seyfarth's separate March 2026 review of federal dockets found 3,117 website suits in 2025, a 27% increase on 2,452 in 2024, making up 36% of all federal ADA Title III filings that year (Seyfarth). One caveat worth holding onto: those counts are federal court only. UsableNet, which also pulls state-court cases from New York and California, projects roughly 6,176 digital accessibility filings for the full year 2026 (UsableNet). Different nets, overlapping pond. The direction is the same in both.

What does an AI-drafted complaint look like?

It reads competent until you check the citations. Seyfarth's lawyers listed the tells they now see in these filings: citations to cases that do not exist, complete with invented parenthetical summaries; descriptions of case holdings that are simply wrong; substantive briefs "written" in less time than it would take a person to type the document; and work product that does not match the plaintiff's spoken command of English (Seyfarth ADA Title III). Those are the fingerprints of a large language model that hallucinated its authority.

Here is the trap in reading too much comfort into that. The fabricated part is usually the legal scaffolding, not the accessibility allegation. A complaint can cite a case that was never decided and still correctly point out that your email field has no programmatic label. The barrier it names is often real, sitting in your markup right now, findable by any scanner. Betting your defense on the plaintiff's AI making a mistake is a bad trade.

Does an AI-drafted complaint change your legal exposure?

Not in the way that matters. The drafting method does not change the underlying question a court asks: can a person using a screen reader or a keyboard actually use your site. My honest read, after watching this trend for a year, is that the AI angle is a distraction from the boring fact underneath it. The complaints keep naming the same handful of machine-detectable failures because those failures keep being there.

Walk through what a template complaint typically alleges and you get a short, familiar list:

  • An <input> with no associated <label> or aria-label, so a screen reader announces "edit text" with no clue what to type.
  • Images that convey meaning with an empty or missing alt attribute.
  • Links whose only content is an icon, giving the accessible name "link" and nothing else.
  • A keyboard focus order that skips the "Add to cart" button or traps focus inside a modal.

None of that requires a lawyer to spot, and none of it requires AI either. It requires that nobody ran the page against WCAG. That is the exposure. The filing tool is incidental.

Who is getting these complaints, and for what?

Mostly smaller e-commerce businesses. UsableNet's 2026 midyear data puts 79% of filings on online stores, and 68% of sued businesses report under $50M in annual revenue (UsableNet). The defendants are, in large part, the companies least likely to have ever run an accessibility scan, not the ones with a compliance team on staff.

The venue picture is concentrated too. In federal court, New York led with 1,021 website filings in 2025, Florida had 961, and Illinois 585 (Seyfarth). California barely shows up in the federal numbers because its plaintiffs file under the state's Unruh Act in state court instead. Where a suit lands is a venue strategy, not a map of where inaccessible sites are.

What actually lowers your risk?

Meeting WCAG on the pages people use, and keeping evidence that you are working at it. The ADA does not name a specific web regulation for private businesses, but the DOJ and the courts treat WCAG as the practical benchmark, so that is the target to build toward (ADA.gov). A workable order of operations:

  1. Scan to find the machine-detectable barriers first. These are the exact issues an AI-drafted complaint tends to cite, and they are the fastest to find and fix. Run your templates and key pages against WCAG 2.2 AA with a free accessibility scan so you have the list with the selector for each finding.
  2. Fix the flows people complete. A checkout or signup form a keyboard user can't finish outranks a hundred decorative-image warnings. Triage by user impact, not raw issue count.
  3. Get a human on the rest. Automated testing catches roughly 30 to 57% of issues depending on how you count (Deque). Whether an error message makes sense through a screen reader is a judgment a person has to make. Federal contractors face the same split under Section 508.
  4. Document the effort. A dated accessibility statement and a saved history of scan reports and fixes is what your counsel wants if a demand letter arrives. It records efforts, not a guarantee, and that is exactly what it should do.

One thing it is worth being blunt about: overlays do not close this gap. Suits against sites already running an accessibility widget keep rising, and in January 2025 the FTC ordered the overlay vendor accessiBe to pay $1 million over claims that its product would automatically make sites compliant with WCAG (FTC). A widget sits on top of your markup at runtime; it cannot rewrite the code a screen reader reads. If you want the criteria laid out by level, the WCAG checklist is the working reference, and the 2026 lawsuit trends post has the fuller filing picture.

FAQ

Are AI tools causing more ADA accessibility lawsuits?

They are lowering the cost of filing, which raises the volume. Seyfarth counted 40% more federal pro se ADA Title III lawsuits in 2025 than 2024 on a monthly-average basis and attributes much of that to AI-assisted complaint drafting (Seyfarth ADA Title III). The law itself hasn't changed; the number of people able to produce a filing has.

How can you tell a complaint was drafted with AI?

The common signs are citations to cases that don't exist, wrong descriptions of what real cases held, and briefs produced faster than anyone could type, sometimes in English that doesn't match the plaintiff's own (Seyfarth ADA Title III). Those flaws are usually in the legal argument, not the accessibility allegation.

Does an AI-drafted complaint make the lawsuit weaker?

Sometimes on the legal scaffolding, rarely on the substance. A hallucinated citation can hurt a specific filing, but the barrier it alleges, like an unlabeled form field, is often real and sitting in your code. Fixing the barrier is a more reliable defense than hoping the plaintiff's AI slipped.

Do accessibility overlays stop these complaints?

No. Filings against sites already running an overlay keep rising, and the FTC fined overlay vendor accessiBe $1 million in 2025 over claims it made sites automatically compliant (FTC). An overlay can't rewrite the underlying markup.

Does the ADA require a specific version of WCAG for private websites?

Not by regulation. There is no dated Title III web rule for private business, but the DOJ and courts treat WCAG as the practical benchmark, so building toward WCAG 2.2 AA is the defensible target (ADA.gov).

See what a complaint would find

The tools drafting these complaints are new; the barriers they name are old. A scan shows you which WCAG issues are on your pages today, ranked so you fix the ones that block real users first, with the exact selector for each finding. Run a free accessibility scan and start with the machine-detectable failures an AI-drafted complaint would flag.

Written by Pavel Charkasau, founder of wcagc.com. I read the Seyfarth filings on the pro se surge so you can plan around what the data says, not the AI hype around it.

Last updated: July 28, 2026

Sources