AI and ADA Website Lawsuits: Pro Se Filings Up 40%

For most of the last decade, ADA website lawsuits followed a predictable pattern. A small group of plaintiff firms filed most of them, working from a small pool of repeat plaintiffs, and the complaints were close to templated. If you were not a consumer-facing store in New York or Florida, the odds of being named felt remote.

That pattern is changing, and the clearest evidence is in who is filing.

Seyfarth Shaw, which tracks federal ADA Title III filings, reported in October 2025 that pro se filings were surging, likely powered by AI. "Pro se" means a plaintiff representing themselves, with no attorney. Their numbers: pro se plaintiffs filed 1,774 federal ADA Title III lawsuits in all of 2024, against 1,867 in the first nine months of 2025 alone. On a monthly average that is a 40% increase. Federal pro se Fair Housing Act filings rose 69% over the same period, which is the same pattern in a different statute.

Two things are worth being precise about, because they are easy to get wrong. That 40% is the growth in pro se filings, not the share of all filings that are pro se, and not a measure of how "easy" suing has become. And ADA Title III covers physical places of public accommodation as well as websites, so the count is not website cases alone. What the figure does show is a sharp change in who brings these cases without a lawyer.

Seyfarth's own explanation for the surge is generative AI. If you build or maintain client sites, that changes which sites are worth worrying about and what a defensible record looks like. Here is what is different, and what it means for the work.

Why a self-represented plaintiff can now file

The mechanics are not complicated. An LLM can read a page, describe its WCAG failures, and draft a complaint that cites statutes and technical standards in the right form. Free automated scanners, the same category of tool plaintiff firms have used for years, supply the findings. Neither step needs a lawyer or a developer.

The cost of actually filing is small. A federal civil action carries a $350 statutory fee under 28 U.S.C. 1914 plus a $55 administrative fee (District Court Miscellaneous Fee Schedule). A plaintiff granted in forma pauperis status pays neither.

The uncomfortable part for defendants is that the underlying findings are often correct. Missing alt text is missing whether the complaint was drafted by a law firm or a chatbot. The accessibility failures a complaint describes are the same failures an automated accessibility scan would report on the same page.

Courts are not ignoring the AI part

Judges have started responding directly. Judge Christopher A. Boyko of the Northern District of Ohio issued a standing order on the use of generative AI that is about as blunt as these get:

"Pursuant to the Court's inherent authority and the authority of Rule 11 of the Federal Rules of Civil Procedure, no attorney for a party, or a pro se party, may use Artificial Intelligence (AI) in the preparation of any filing submitted to the Court."

The order lists possible sanctions including striking the pleading from the record, economic sanctions or contempt, and dismissal of the lawsuit. It carves out legal search engines such as Westlaw and LexisNexis and general search engines such as Google and Bing, and it puts a duty on parties and counsel to tell the court if they discover AI use in their case.

That is one judge's standing order in one district, not a national rule. But it is a concrete sign that an AI-drafted complaint carries its own risk for the person filing it. If you are on the receiving end of one, verifying that every case it cites actually exists is a reasonable first instruction to your lawyer.

Small sites are not out of range

The old reasoning was that lawsuits chase deep pockets, so a small store stays under the radar. The filing data does not support it. UsableNet's 2026 review of ADA web lawsuit trends reports that 64% of companies sued generate less than $25 million in annual revenue, and that more than 5,000 digital accessibility lawsuits had been filed by the end of 2025.

Geography is still concentrated but no longer decisive. UsableNet reports that New York alone accounted for well over a third of state-level ADA website lawsuits, with nearly 2,000 cases filed in New York and California state courts combined and more than 3,100 federal filings on top. Courts have repeatedly held that a defendant's physical location is not what decides whether a website case can be heard.

For an agency, the practical read is that "this client is too small to be a target" is not a filter you can apply to a portfolio any more.

What one court accepted as evidence

There is a more encouraging development, and it is worth knowing precisely.

Seyfarth Shaw reported in April 2026 that Judge J. Paul Oetken of the Southern District of New York dismissed a website accessibility case as moot on unrebutted evidence that the site was accessible. The defendant submitted a declaration from its Director of E-Commerce describing work with a digital accessibility consultant since 2023, the investigation of the specific barriers alleged, a low-severity issue that had since been remediated, and the consultant's own report showing the audit process and findings. The declaration described "commercially reasonable" steps and ongoing work. The plaintiff did not rebut it.

Read that for what it is: one dismissal, on that record, in that district. It is not a rule that documentation defeats a claim, and nothing you buy or run makes a site lawsuit proof. What it does show is the kind of record a defendant was able to put in front of a court: dated findings, specific remediation tied to those findings, and work that predates the complaint rather than starting after it.

That is also the opposite of an accessibility widget. The FTC approved a final order in April 2025 requiring accessiBe to pay $1 million and barred the company from claiming its automated product makes or keeps a website WCAG compliant without supporting evidence. A widget is not a remediation record.

What this means for the work

If accessibility has been sitting behind other work because the risk felt theoretical, the filing numbers are a reason to move it up. None of this is legal advice, and an agency should say so when it comes up with clients.

• Scan the pages that matter, and keep the output. The failures that show up most often in these complaints are missing alt text, low contrast, unlabeled form fields, keyboard traps and missing focus indicators. Start with checkout, product pages and forms. Our form labels checker and keyboard and focus checker explain what each of those checks looks at.
• Date everything. A dated scan, the issues it found, and the commit or ticket that fixed each one is the paper trail. Undated work is hard to show anyone later.
• Rescan on a schedule rather than auditing once. Sites change with every theme update, new template and third-party script, and accessibility regresses quietly.
• Fix the failures that block people first. A 500-item report is paralyzing. Errors that stop someone completing a task come before best-practice notices. Our accessibility audit checklist sets out the order.
• Be careful with overlays. If a client has one, read what overlays do and do not do for lawsuit risk before renewing it.
• Write an accessibility statement that matches reality. What has been done, when, and how a user reports a problem, rather than a template.

The tooling cuts both ways

The same generative AI that lowers the cost of drafting a complaint also lowers the cost of fixing what the complaint is about. A scanner can find the machine-detectable failures and produce the markup that fixes them, which is work that used to need a developer and a consultant.

What has not changed is the limit on automation. Automated checks find many real problems and cannot confirm conformance: alt text quality, focus order in a real flow and whether an error message actually helps still need a person. AccessGuard runs automated WCAG 2.2 AA checks in a real browser, keeps every client site in one account with new and fixed failures since the last scan, and produces white-label PDF reports under your agency's name. On paid plans it writes the fixes and rescans on a schedule you set, so the record builds itself.

If you want to see where a client site stands today, start with a free accessibility scan, then read the WCAG issues behind ADA website lawsuits for what to fix first.