Towards Radical Approaches for Web Accessibility

ASSETS 2026 – Workshop

Envisioning Strategies and Interventions in a Shifting Web

The Web is shifting again: increasingly built with AI tools by people without accessibility expertise, and increasingly optimised for AI agents over human readers. As with every previous shift, this risks leaving disabled people further behind. This workshop asks what we should do about it.

At a glance

Submit by: 19 August 2026 (anywhere on Earth)
Notifications: 24 August 2026
Format: 2–4 page position paper, provocation, pictorial, or podcast
Send to: luis.a.p.carvalho@northumbria.ac.uk
Attend: at least one author per accepted submission; all participants register for ASSETS 2026

Call For Participation

The Web remains largely inaccessible. Two decades of guidelines have not, on their own, produced a healthy and accessible Web, which tells us that current and past approaches have not been enough. We believe the HCI community now needs radical approaches that go beyond guidelines: approaches equal to the Web's growing complexity, and capable of supporting global strategies for tackling the accessibility crisis.

This workshop brings together researchers, developers, and industry and policy practitioners to explore those approaches through short presentations, brainstorming, and group discussion. It is a first step towards reframing web accessibility for a Web that is changing beneath us.

To take part, prepare a short (2–4 page) position paper, provocation, pictorial, or podcast on a novel approach to web accessibility. Suggested topics include, but are not limited to:

Intervention scale: global, national, regional, local
Concept borrowing from other fields
Monitoring and predicting for accessible web futures
Web accessibility in the Age of AI

How the review works. Submissions are read in two rounds. First, we assess relevance to web accessibility and how novel and provocative the approach is. Then prospective participants read the shortlist and rank the submissions they would most like to discuss on the day. Submissions are anonymised before this second round, so please do not identify yourself in the document. Accepted submissions are made public and hosted here.

Send your submission by email to Luis Carvalho at luis.a.p.carvalho@northumbria.ac.uk.

Themes

To prompt this exploration, the workshop is organised around four themes.

Theme 1 – Intervention Scale: Global, National, Regional

Who, beyond individual site owners, should be responsible for an accessible Web?

Web accessibility has historically relied on the goodwill of developers and companies. In recent years, particularly in the public sector, legislation such as the EU Web Accessibility Directive and Section 508 in the USA has begun to hold organisations accountable. But the same laws often contain clauses, such as "disproportionate burden", that let accessibility be de-prioritised, and they risk turning accessibility from an ethical commitment into a defensive response to the threat of litigation. In this theme, we explore the future of web accessibility across every scale of society, and the roles other stakeholders might play.

Theme 2 – Concept Borrowing

What can other fields teach web accessibility?

Borrowing concepts and methods from other fields can generate genuinely new thinking. In mobile app accessibility, Ross and colleagues built a framework inspired by epidemiology, in which a barrier behaves like a disease: it infects a host (the app) via an infectious agent (a UI component) carrying a determinant (the root cause), and spreads through a transmission mechanism (a UI library). It treats accessibility as the product of a system's interaction with its environment rather than something to study in isolation. In this theme, we invite participants to bring the ideas and methods of other fields into web accessibility.

Theme 3 – Monitoring and Predicting for Accessible Web Future

Can we catch barriers before they spread, rather than measure them after the fact?

Today's tools mostly assess sites after they are built and deployed, and the Web's dynamic nature means those measurements date quickly. In this theme, we look beyond compliance scoring towards predictive approaches that draw on trends in web development to anticipate change, and towards tools that prevent barriers before they permeate the Web.

Theme 4 – Web Accessibility in the Age of AI

What does accessibility mean when the Web is built by, and for, AI?

AI has served accessibility for years, from teachable AI that helps blind and low-vision people find their belongings to captioning and image description on the Web, with accuracy the persistent limitation. Generative AI now puts real capability in the public's hands, opening the door to citizen-led and grassroots tools, with early work already using large language models to detect and fix barriers on live pages. At the same time, the era of freely scraping the Web is ending, and what replaces it could make large-scale accessibility evaluation impossible to run. In this theme, we invite approaches that empower ordinary users to shape the Web's accessibility, and that keep large-scale evaluation viable.

Workshop Goals

This workshop aims to bring together researchers, developers, and industry members who want to push web accessibility beyond guideline assessment. Our goals are to:

Identify new approaches for tackling current and future inaccessibility on the Web, including the gaps they address and the challenges they face.
Propose a new view of web accessibility through a blue-skies exploration of what the field could become, and what its objectives and global strategies should be.
Revitalise web accessibility as a research field by building a lasting community of researchers, developers, industry members, and policymakers who continue this work beyond the workshop.

Background & Motivation

We are at the edge of another shift in how the Web is built. More and more, websites are produced with AI tools, not only by developers but by people with no background in web development or accessibility. In many ways this democratises the Web. For accessibility, though, it risks producing inaccessible websites at scale.

Shifts like this recur throughout the Web's history. From dynamic content to mobile and responsive design, each was driven by what developers and non-disabled users wanted, while accessibility was never on the default path. The assumptions accessibility depended on broke at each shift, and accessibility was retrofitted only after new barriers were in place. When dynamic pages began updating without a full reload, for instance, they broke the page-load model assistive technologies relied on, and WAI-ARIA had to be introduced to restore the semantics that HTML alone could no longer convey.

A second shift is now under way, in who the Web is built for. Its content has always been prioritised for some audiences over others, with disabled people typically coming second, after the "default" non-disabled user. A new audience now sits above both: the AI agent. As crawling, summarising, and answering become the dominant way the Web is used, content is increasingly optimised for machines rather than human readers.

This arrives at a difficult moment. Despite twenty years of guidelines, study after study still finds sites failing even basic accessibility checks, and the annual WebAIM Million reports continue to find that the overwhelming majority of the top million home pages contain at least one detectable barrier. Tellingly, the improvements that have been found are usually a side effect of new web technologies, not better guideline adoption.

These explanations rest on assumptions the emerging Web may no longer meet: that there is a knowledgeable human in the development loop who can be educated and held accountable; that content is authored for human consumption; and that technology shifts might move in accessibility's favour. In a Web built by and for AI, all three could break, starting another cycle of retrofitting. We therefore argue for radical approaches beyond guidelines and education, and for large-scale interventions that address the websites that already exist. Although our focus is the Web, we intend the discussion to surface strategies that transfer well beyond it.

Indicative Schedule

The workshop runs for four hours, with short breaks throughout:

Introduction and welcome (10 min)
Submission presentations: selected papers, provocations, pictorials, and podcasts, up to 10 minutes each (up to 40 min)
Break (5 min)
Novel approaches for web accessibility: theme-based groups brainstorm challenges, knowledge gaps, and connections to their own work, then rate approaches for feasibility and compatibility (50 min)
Longer break (30 min)
Discussion: each group presents and we share feedback across the room (45 min)
Break (5 min)
Envisioning a new web accessibility, and moving forward: defining the core concepts and global strategies for a step-change in the field (45 min)
Reflections and wrap-up (10 min)

Organisers

Luis Carvalho is a PhD student at Northumbria University. His PhD research focus on exploring the root causes of web accessibility barriers through an epidemiology lenses, with particular focus on the role third-party technologies play on the propagation of accessibility barriers in the web. Previously, he worked as a research assistant on the "Cue Band" project which aimed to provide an non-invasive treatment for diminishing drooling symptoms on people with Parkinson's via wearable technology and on the project "Enabling Ongoingness: Content Creation & Consumption in the New Digital Age" which explored how how we can create meaningful connections with people who have passed on or who are at the end of life through bespoke technology.
Christopher Jackson is a PhD student in Northumbria University's CDT in Citizen-Centred Artificial Intelligence. His research co-designs accessible workshops for students living with energy-limiting conditions, drawing on crip theory and crip technoscience to explore how AI tools can help these students shape their university experience. Chris previously spent over a decade in higher education, in professional support roles helping students at all levels.
Anne Spencer Ross is an Assistant Professor at Bucknell University. Building on her epidemiology-inspired framework for digital accessibility, she is researching a range of social, technical, and organisational factors that impact software's accessibility. Her current research includes how small businesses and organisations manage their websites and social media. Positioned at an undergraduate, teaching-focused university, she is also exploring integrating accessibility concepts into computing curriculum.
Tiago Guerreiro is a Professor of Computer Science at Faculdade de Ciências da Universidade de Lisboa and a researcher at LASIGE where he leads the Tech&People Lab. His research is focused on improving accessibility and health, by working closely with the communities he works with, co-designing solutions, developing and assessing their impact in the wild, in carefully-designed deployments. He is Editor-in-Chief for ACM Transactions on Accessible Computing and was ASSETS 2020 General chair. He was the Web for All General and Program Chair, in 2016 and 2015, respectively, was Chair of the “Accessibility & Aging” Subcommittee for CHI 2019, among many other service roles. He was an invited expert supporting the European Commission in implementing the Web Accessibility Directive.
Kyle Montague is a Professor at Northumbria University. His research spans many topics and domains with a single unifying vision: to address critical social problems by designing and configuring digital technologies that empower individuals and marginalised communities. In particular, his work seeks to democratise access to the tools and processes by which we build the technologies and services that shape society.

Accessibility

This workshop runs entirely in person, and we want everyone to be able to take part fully. Ahead of the day we will ask participants about any requirements or adjustments they need, so the specifics may evolve as we hear from you. If you have an access requirement, please let us know by emailing Luis Carvalho at luis.a.p.carvalho@northumbria.ac.uk.