Forbes Just Ranked Accessibility's Best. The Rest of the Web Got Worse

Steffy McQuiggin is deaf-blind. She holds a conversation using a robotic hand built by Tatum Robotics, called Tatum1, which turns spoken and written words into sign language she can feel through her fingers. It's one of the standout stories in Forbes' 2026 Accessibility 200, published 19 May and expanded for its second year from 100 honourees to 200, spanning 23 countries and six continents, and recognising orgnaisations for accessibility innovation and impact rather than for verified compliance with technical standards such as the Web Content Accessibility Guidelines (WCAG).

This year's list leans heavily on artificial intelligence as the thread connecting its honourees. Amazon's "Help Me Decide" tool narrows an overwhelming product search down to two or three options based on past behaviour, aimed at users with limited vision or certain forms of autism who find infinite scroll genuinely disabling rather than merely annoying. Its Prime Video Dialogue Boost feature separates dialogue from sound effects for anyone who struggles to follow speech under a film's score. Microsoft has fed more than 20 million hours of footage from Be My Eyes - the app that connects blind users with sighted volunteers over live video - into the systems behind its own accessibility tools, and works with the University of Illinois' Speech Accessibility Project to help voice AI understand speech affected by stammering, Parkinson's, or deafness.

None of this is invented for the awards cycle. Tatum's founder, Samantha Johnson, described the real difficulty of the work: building AI on sign language datasets that mostly don't exist yet, because ASL isn't English rendered differently - it has its own grammar, and a lot of it was never digitised in the first place. Microsoft's Neil Barnett puts the upside plainly: AI is making it easier to build accessible products "out of the gate," rather than retrofitting them once someone complains.

That's a very different story to the one most organisations are living through this year. The web didn't get more accessible in 2026, in fact it got measurably less so. WebAIM's February analysis of the top one million homepages found 95.9% carrying at least one detectable WCAG failure, up from 94.8% the year before: the first regression in six years of tracking. UsableNet's midyear lawsuit data has 2026 on track to exceed 6,000 digital accessibility filings in the US, roughly 20% above 2025's already-record total. And Level Access's latest survey found 92% of organisations at least "somewhat confident" in their own accessibility, while close to six in ten believe they could still face litigation or regulatory action over it. Confidence and reality have quietly come apart.

Which is really the point worth taking from this year's list. The Accessibility 200 measures something real - whether an organisation is building something that visibly, measurably changes a disabled person's life, judged by breadth of impact rather than niche excellence. Internal programmes don't count unless their impact reaches beyond the company's own walls; advocacy work and charitable giving are excluded almost entirely, because the list is built to recognise innovation and reach, not intention. It's curated from open nominations, editor research, and a twelve-member advisory board -not a technical audit, and not trying to be one.

That distinction matters this year specifically, because AI personalisation and WCAG compliance are solving different problems. A voice assistant that adapts to someone's speech pattern, or a shopping tool that narrows an overwhelming page down to three good options, can be an excellent, honouree-worthy piece of work, however none of it tells you whether the same company's checkout flow works with a screen reader, or whether a keyboard-only user can complete an ordinary task without hitting a dead end. Personalisation happens inside a product that already works. It doesn't make a broken one accessible.

For anyone building or buying digital products rather than shopping for AI features, the more useful habit is smaller and less headline-friendly than anything on the list: ask who can't use this before it ships, not after someone complains. Get an audit against WCAG 2.1 or 2.2 AA on the actual live product, not a marketing sample. And treat your own accessibility statement as evidence rather than decoration as under the enforcement now underway across the EU and the US, a statement that claims more than the product delivers is a liability in its own right, not a shield.

Recognition and readiness are both worth having. Only one of them holds up when someone actually tries to use what you've built. Forbes will publish next year's list on its own timetable. Whether your own product or service would survive an actual accessibility audit is a question you can start answering today.