A Federal Judge Just Turned Web Accessibility Into a Design Spec, Not a Checkbox
Most accessibility fixes ship the way most bug fixes ship. Someone files a complaint about one broken thing — a form that doesn't announce errors to a screen reader, a modal that traps keyboard focus — and a designer patches that one thing. The complaint closes. The rest of the site, built on the same patterns, stays exactly as inaccessible as it was before. That model just got a lot harder to hide behind.
In July 2026, Judge Haywood S. Gilliam Jr. certified a nationwide Rule 23(b)(2) injunctive class in Merrell v. Ralph Lauren Corp. in the Northern District of California. The reasoning is the part that should actually change how design teams work, not just how legal teams respond. The court held that a single injunction requiring ADA-compliant remediation would provide relief to every member of the class. Its logic: the barriers weren't scattered defects. They were properties of the site's underlying architecture. You don't get to fix the barrier one plaintiff hit and call the system compliant — the system produced that barrier for everyone standing in a materially similar position, whether or not they'd filed a complaint yet.
The Ruling Treats a Website as One Object, Not a Pile of Features
This is a genuinely different framing than the one most design orgs operate under. Component-level accessibility audits — check the button contrast here, add an aria-label there — treat a site as a collection of independently fixable parts. Gilliam's ruling treats it as a single, indivisible navigability system. If the information architecture, form layout, and interaction patterns produce a barrier in one instance, the same architecture produces the same barrier everywhere it repeats — and in a component-based design system, that's almost everywhere. The court also rejected Ralph Lauren's attempt to carve the case down along an "intent to transact" line, the argument that browsing without an active intent to purchase shouldn't count toward the injury. Instead it held baseline navigability to the same standard regardless of what the visitor was there to do. That closes off a defense a lot of legal teams have leaned on for years.
The practical shift for design teams: the unit of compliance is no longer "did we fix this specific reported issue." It's "does the pattern that produced this issue exist anywhere else in the system." Under a design-system architecture, the honest answer to that question is almost always yes.
Design Systems Made This Both Easier and More Dangerous
There's a genuine irony sitting underneath this ruling that design leadership should sit with. Component libraries and design systems were supposed to make accessibility easier to enforce — fix the pattern once in the source component, and the fix propagates everywhere the component is used. That's true. It's also exactly why an unfixed accessibility defect in a shared pattern is now more legally dangerous than it would have been in a site built from one-off, unrelated pages. A bespoke page-by-page site produces isolated barriers a plaintiff has to prove individually. A design-system-built site produces the same barrier at scale, consistently, by construction. That's precisely the fact pattern that makes a class-wide injunction easy for a court to justify. The very consistency that makes a design system valuable is what turns one component-level oversight into a sitewide liability, the moment a court decides to look at the whole system instead of the one page someone happened to sue over.
This mirrors the accountability shift already underway in AI agent design, where regulators increasingly ask not who wrote the code that caused harm, but who designed the system-level pattern that made the harm structurally likely. Accessibility litigation is arriving at the same conclusion from a different direction. The design decision, not the individual instance, is where the liability actually lives.
What an Audit Looks Like When the System Is the Defendant
A remediation plan built for this legal environment looks different from a punch-list of reported issues. It starts by identifying the design-system-level patterns — not the individual pages — that produce noncompliant behavior. Those are the patterns doing the damage at scale, repeated across every page that uses them. It requires design and engineering to treat a WCAG failure in a shared component as equivalent in urgency to a failure on the highest-traffic page on the site, because under a class-wide framing, they carry the same legal weight. And it means accessibility can no longer live downstream of design review, as a compliance pass that happens after visual and interaction decisions are locked. A pattern approved at the component level without an accessibility check is now a liability approved at the systemic level — instantly, the moment it ships into production.
So Actually, the Checkbox Was Always the Wrong Mental Model
Treating accessibility as a checklist implied it could be completed — audit the site, fix the flagged items, move on until the next audit cycle. Merrell is a court explicitly rejecting that model in favor of a harder, more accurate one. Accessibility is a property of the architecture itself, continuously produced or violated by every pattern that architecture repeats. It isn't a state you achieve and then maintain by inspection.
The teams that treat this ruling as a one-time remediation sprint will pass the specific audit in front of them and remain exposed to the next one. The teams that treat it as what it actually is will not. This is a court confirming that design-system patterns are now a standing legal surface, not a design preference. Stop generating the same liability with every new component you ship.