How AI industrialized accessibility lawsuits: The crawl-by era

Automated web crawlers have replaced physical plaintiffs. In the first half of 2025, the U.S. saw 2,014 digital-specific accessibility lawsuits—a 37% year-over-year increase. That's not because compliance got harder. It's because the barrier to entry collapsed.

Ten years ago, an accessibility lawsuit required a plaintiff to physically visit a location, document violations in person, and hire a lawyer. Today, an automated script can crawl thousands of e-commerce sites in hours, flag machine-detectable failures, and auto-populate a draft complaint. The cost of entry dropped from thousands of dollars to zero.

This is the crawl-by lawsuit era. And it's driven entirely by AI and automation.

How the crawl-by replaced the drive-by

The ADA's Title III (which covers public accommodations, including websites) originally assumed plaintiffs would physically test locations. A person would visit a store, encounter barriers, document them, and sue. This friction limited volume. Litigation remained limited in scale.

That model broke down over the last five years. Web accessibility lawsuits exploded because the internet scales. One plaintiff can claim injuries across thousands of sites. One lawyer can file 40 cases per month using templated complaints. Automation industrialized what was once a manual process.

Here's the workflow:

Step 1: Automated scanning. A tool crawls a website and detects machine-readable WCAG failures. Missing alt text (`WCAG 1.1.1`), empty form labels (`WCAG 3.3.2`), low contrast text (`WCAG 1.4.3`). The scanner generates a report in minutes.

Step 2: Complaint auto-population. Law firms and platforms feed scanner results into templates. A paralegal (or AI model) fills in the boilerplate. The complainant alleges they visited the site, encountered barriers, and were injured. The specific violations are listed verbatim from the scan report.

Step 3: Settlement pressure. The defendant receives a demand letter. Fighting typically costs

5K–$50K in legal fees (per Seyfarth Shaw litigation data). Settling typically costs
0K–
5K. The math is brutal. Most companies pay.

Step 4: Repeat. Same script, different domain. Scale matters. Volume lawsuits work because individual settlements are predictable.

This is not litigation. It's a process. And processes get automated.

The pro se revolution: AI as a legal tech platform

Pro se litigants—people representing themselves without lawyers—now file 40% of all ADA Title III digital accessibility cases. That percentage has climbed sharply since 2023, when generative AI tools became widely available.

Why? Because ChatGPT and legal-tech platforms have democratized complaint writing.

A person no longer needs a lawyer to draft a complaint. They can describe a violation in plain English, ask an AI to convert it into legal prose, upload a screenshot from an accessibility scanner, and file the complaint themselves. They're not getting legal advice. They're getting a secretary.

The economics are staggering. A lawyer charges

00–$400 per hour. Drafting a complaint takes five to ten hours if done carefully. That's
,000–$4,000 in legal fees before filing. ChatGPT and other generative AI tools do it in minutes for $0.

Some pro se plaintiffs are genuinely motivated by accessibility. Others are running a numbers game. The difference barely matters. Both file complaints. Both settle cheap. The only difference is who pockets the settlement.

Legal-tech platforms have amplified this. Services that let non-lawyers file complaints with minimal friction, combined with an AI integration that auto-populates violations from a scan report, have turned accessibility litigation into a gig-economy activity. A person with a few hours per week can file five or ten complaints per month and collect settlement checks.

Target selection: who gets sued and why

AI accessibility scanners have become the prosecutor, judge, and jury. They decide which sites get sued, why, and in what order.

Here's the catch: automated scanners detect approximately 20–30% of actual WCAG criteria. They're good at finding missing alt text and empty form fields. They're terrible at assessing real-world accessibility—whether a visually impaired person can actually complete a checkout, whether a deaf person can access video content, whether a person with dyslexia can read the interface.

This creates a systematic bias: lawsuits disproportionately target the 20–30% of non-conformities that machines can detect. They ignore the 70–80% that require human judgment.

The result is predictable. Certain industries and site types get hammered. Others get ignored.

What to do this week

You can't stop automated crawling. But you can reduce your profile.

Audit your site using the same tools your litigants use. Use tools like inspekter, WAVE, or Lighthouse to run an automated scan. Look for the wins: missing alt text, empty form labels, low contrast ratios, missing button text. These are the 20–30% of non-conformities that automated tools catch—and that automated lawsuits target. Fix them.

Prioritize e-commerce and checkout flows. If you sell online, this is where the risk concentrates. Audit every form field, every button, every error message. If your site is Shopify, audit harder. The risk premium is real.

Test with real users, not just tools. Automated scanners miss 70% of actual accessibility issues. A person using a screen reader will find problems that a tool never flags. Spend two hours with a real user and you'll understand your site better than 50 automated scans.

Document your efforts. If you do get sued, showing evidence of genuine remediation efforts—not just automated fixes—strengthens your defense. A lawyer can argue that you're making good-faith improvements, not ignoring the problem.

Know your revenue threshold. If you're under

5M in revenue, you're statistically more likely to be targeted. The cost-benefit of litigation tilts toward defendants at higher revenue scales. That's not fair, but it's real.

The series continues

This is Part 1 of "AI and the accessibility lawsuit machine." In Part 2, we'll examine what happens when AI-generated legal filings contain fabricated case law—and why that still won't save your website.