I Scanned 42 US Government Websites Two Weeks Before the ADA Title II Deadline

Aleksandr Moon — Wed, 08 Apr 2026 21:49:43 +0000

What 42 Government Websites Actually Look Like

On April 24, 2026, a piece of the ADA that most people never thought about becomes enforceable. State and local governments must make their websites and mobile apps accessible using WCAG 2.1 Level AA standards. You have 16 days left if you live in a place with more than 50,000 people.

I wanted to know how ready the states are. Not through abstract concepts, self-reported data, or press releases claiming commitment to accessibility. I wanted numbers I could generate myself in an afternoon with a free tool so anyone reading this could reproduce them. I found states where I could access both the executive and judicial branch websites, then tested each one with axe-core. Twenty-six pairs loaded from both sides. I also chose twenty major US cities because Title II applies to local governments and most Americans use city websites to access public services.

Three of my colleagues from the judicial system told me that courts are performing worse. Smaller IT budgets, older CMS installations, fewer compliance staff. "Check the courts" was the exact phrase I kept hearing.

That was half right and half wrong.

Quick primer on what's at stake

The Americans with Disabilities Act became law in 1990. Title II has always required state and local government services to be accessible. What it didn't spell out was a specific technical standard for websites. In April 2024, the Department of Justice published a final rule requiring public entities to follow WCAG 2.1 Level AA standards. Large governments serving 50,000 or more people must comply by April 24, 2026. Smaller entities get another year. Neither date moves.

Enforcement comes from three directions. The DOJ can investigate, negotiate settlements, and file lawsuits. People and advocacy groups can file federal civil rights complaints or sue under the ADA. Courts can order remediation, award attorney's fees, and in some cases compensatory damages. Most cases end in settlements where public entities must hire outside auditors, fix problems, and report progress for years instead of going to court. Not cheap, and embarrassing in a way that lingers in local press coverage.

Here's the detail that matters most for the rest of this piece. When someone files a federal complaint about website accessibility, the state government is responsible — not the courts, the vendor, or the contracting department. The public entity is the legal unit. What happens next depends on what the rest of that state's digital footprint looks like, which is the part I wanted to measure.

The 26-state table

Each row is a single homepage scan, default axe-core ruleset, letter grade from a severity-weighted score.

State	Executive	Judicial
Alabama	B (83) / 14	C (78) / 18
Alaska	D (50) / 136	A (96) / 3
Arizona	C (72) / 19	F (44) / 47
Arkansas	B (85) / 49	B (87) / 10
California	A (100) / 0	A (94) / 8
Colorado	A (100) / 0	A (100) / 0
Connecticut	A (92) / 6	B (87) / 22
Indiana	A (97) / 2	A (97) / 2
Kansas	B (89) / 20	A (90) / 14
Kentucky	D (51) / 58	B (86) / 12
Maine	B (80) / 38	B (87) / 19
Maryland	A (100) / 0	B (88) / 7
Minnesota	B (89) / 15	D (56) / 36
Mississippi	D (58) / 39	A (94) / 4
Montana	A (96) / 3	B (86) / 35
New Jersey	B (89) / 5	A (95) / 3
New Mexico	C (73) / 16	B (90) / 6
North Dakota	B (90) / 11	A (96) / 3
Oklahoma	A (98) / 2	B (82) / 22
Rhode Island	C (70) / 34	C (77) / 17
South Carolina	A (92) / 5	A (99) / 1
South Dakota	A (100) / 1	A (100) / 0
Tennessee	A (95) / 3	B (88) / 13
Utah	A (95) / 4	D (55) / 42
Virginia	B (88) / 17	B (85) / 9
Wisconsin	B (87) / 11	A (96) / 4

What I expected vs what I found

I expected the bolded rows to be a column on the judicial side. I thought courts would be worse and the executive branch would be cleaner.

Arizona follows this pattern, with arizona.gov getting a C grade and azcourts.gov getting an F — the only F grade among all 26 states studied. Minnesota fits it. So does Utah, the sharpest case in the table: utah.gov scores 95 with four findings, and utcourts.gov scores 55 with forty-two. Same state, same week, same federal deadline, and the judicial branch has ten times as many problems and drops three full letter grades.

Three clean cases of "courts are worse." If the table stopped there, I'd have a clear, critical piece about state court IT budgets.

Then I looked at the other bolded rows.

Alaska's executive branch homepage scores a 50. The page had 136 problems when it loaded, making it the worst executive score in the dataset. In contrast, courts.alaska.gov scored an A with only three problems — a forty-six-point spread in the wrong direction. Kentucky has the same shape: kentucky.gov at D(51) with fifty-eight issues, kycourts.gov at B(86) with twelve. Mississippi's state website scored poorly with a D (58 points, 39 problems), but their courts website scored an A (94 points, 4 problems).

Three states where courts are much worse. Three states where the executive branch is much worse. I pulled the dataset by walking straight down a list of state abbreviations, so there's no cherry-picking on either side. The branch isn't the variable. The branch makes no predictions.

The rest of the table is the actual story

Twenty of the twenty-six states have similar letter grades on both sides. Eighteen of those twenty have both branches at A or B. Colorado and South Dakota score perfect 100s on both halves of state government. Indiana returns A(97) with identical findings on both sides, likely from the same vendor or team running both sites. Rhode Island is the only state where both branches land at C; it performs the same on both sides, symmetrically.

That symmetric cluster is the majority of the data. Twenty out of twenty-six states don't have a branch problem at all. A state either has a working accessibility culture or it doesn't, and whichever it is, this affects the entire state government.

The six outlier states — three with one branch clean and the other not, three with the opposite arrangement — are states where one branch has that culture and the other doesn't. That's not a branch story. That's a fragmentation story. Kentucky's courts team is doing fine and Kentucky's executive team is not. Utah's executive team is doing fine and Utah's courts team is not. One office does the work and the other doesn't, depending entirely on which office got the memo.

Cities

Here are the sixteen big-city sites that loaded.

City	Grade / Issues
New York	A (95) / 2
Los Angeles	A (96) / 3
Chicago	A (90) / 8
Houston	B (90) / 13
Phoenix	A (95) / 3
Philadelphia	B (81) / 59
San Antonio	A (92) / 6
San Diego	B (88) / 11
Austin	A (90) / 24
Fort Worth	B (88) / 14
Columbus	A (100) / 0
Charlotte	A (100) / 0
Indianapolis	A (100) / 0
Seattle	A (98) / 1
Washington, D.C.	C (76) / 46
Boston	F (44) / 58

Most of the top twenty cities are in surprisingly good shape. Columbus, Charlotte, and Indianapolis — three big cities that nobody writes accessibility case studies about — all returned perfect hundred scores with zero automated findings. Five more cities received A grades: NYC, LA, Chicago, Phoenix, and Seattle.

Boston was the only city to receive an F grade out of all 42 sites tested. Forty-four out of a hundred, fifty-eight findings on the homepage of a city that's spent years publicly branding itself as a model of civic digital service. Washington, D.C. at C(76) is the other one worth a second look — dc.gov is the most-visited municipal site in the country, and sixteen days from now it falls under exactly the same rule as every town of fifty thousand people in Kansas.

Why this matters for April 24

I couldn't predict a state or city's Title II readiness just by knowing its name. Not from population. Not from political leaning. Not from branch of government. Not from whether the mayor has given a recent speech about accessibility. Variance within individual states often exceeds variance between different states.

A federal deadline doesn't care which department owns which site. It lands on the whole public entity as one legal unit. But government accessibility teams work separately — different teams maintain minnesota.gov and mncourts.gov, and they probably don't share a contractor, a CMS, a budget line, or a compliance officer. This isn't "governments don't care about accessibility." It's that ownership is fragmented, and in most of the failing cases I found, somebody somewhere in that same government is already doing it right and nobody ever walked next door.

Caveats and what to do

Axe-core catches somewhere between 30% and 40% of real WCAG issues depending on which study you trust. Catching keyboard traps, focus order issues, screen reader announcement quality, and meaningless alt text needs a human with assistive tech and some time. Every number above is a floor, not a ceiling, and the failing rows almost certainly look worse under manual review, not better.

If you own one of these sites, three things. First: audit every branch and department of your organization, not just the main .gov site. If the results look very different, that by itself is a finding — you have fragmented ownership. Second: focus on serious and critical issues first. A site with ten problems including one critical is more urgent in practice than a site with forty moderate ones. Third: don't trust the press release. Run the scan on the production URL today and save the output.

Check mobile first. Most of these sites are clearly optimized for desktop, but mobile versions lose grade points — and in 2026, roughly nine out of ten visitors to a government website are on a phone.

Sixteen days

Not long enough to redesign anything. Long enough to fix the top ten serious problems on every entry-point page if somebody clears a developer's calendar to do it. The cleanest governments in these scans aren't the richest ones — they're Columbus, Indianapolis, Colorado, and South Dakota, places where somebody treated this as a routine line item in the budget, not a PR problem. The governments that look worst aren't the ones with the worst developers; they're the ones where the line between "our team's site" and "their team's site" was drawn too early, and nobody checked the other side of it until a federal deadline forced the whole public entity into the same frame.

Sixteen days is enough time to check the other side.

*All 42 scans were run in a single afternoon using AccessCheck, a free Chrome extension that wraps axe-core in a letter-grade report. Any axe-core-based tool will give you comparable numbers.

DEV Community: Aleksandr Moon