Accessible AI: Who is responsible when AI acts on our behalf?
Robin Christopherson | 02 Sep 2026Most of today’s AI tools wait for us to ask a question. Increasingly, the next generation will help us complete tasks.
We are already seeing early examples of AI systems that can compare options, complete actions, and work across multiple services on a user’s behalf. Over the next few years, that capability is likely to expand rapidly.
As someone who has spent nearly three decades working in accessibility, I find that development fascinating. And as a blind person who has spent a lifetime navigating digital barriers, I also find myself thinking carefully about what it means for independence, control and trust.
For disabled people, agentic AI could open doors that have remained stubbornly closed for years. It could also introduce entirely new challenges if accessibility is treated as an afterthought. That makes this a particularly important moment to ask how these systems should behave, who gets to shape them and how disabled people can remain fully involved in decisions that affect them.
From interfaces to outcomes
For most of the digital age, accessibility work has focused on helping people navigate interfaces. We have worked to improve websites, apps, services and documents so that people can perceive, understand and operate them effectively.
Screen reader support, keyboard access, captions, colour contrast and clear content all remain essential parts of that work. Agentic AI introduces a different dynamic.
Rather than moving step by step through an online service, users may increasingly be able to describe what they want to achieve and ask an AI assistant to help complete the journey. Booking travel, comparing products, managing household bills, renewing insurance or completing administrative tasks could become conversations rather than processes.
The appeal for many disabled people is easy to understand. Much of the friction we encounter online comes from complexity. A task may technically be possible, yet still require more time, effort and persistence than it does for someone who is not navigating with assistive technology or managing another access need. An effective AI agent has the potential to remove some of that burden.
Why this particularly impacts financial services and other complex (and often highly regulated) sectors
Some of the most demanding digital journeys exist within sectors such as banking, insurance, retail, utilities and government services. These are often high-stakes interactions involving detailed information, multiple steps and significant consequences if something goes wrong.
Customers may need to compare products, understand terms and conditions, provide evidence, complete forms and revisit decisions over time. Accessible design has already made many of these journeys considerably better than they once were, but barriers remain.
Agentic AI could help customers navigate those journeys more effectively, particularly where people are managing large amounts of information or interacting with multiple systems. For disabled customers, this could support greater independence.
For organisations, it offers an opportunity to improve customer experience in ways that benefit a much wider audience than disabled people alone. Many accessibility improvements follow that pattern: they begin by addressing specific barriers and ultimately make life easier for everyone.
Visibility matters as much as convenience
The accessibility challenges associated with agentic AI differ from many of the challenges we have dealt with previously. Traditional barriers are often visible and relatively easy to identify: A form field might be unlabelled. A keyboard user may be unable to reach a control. Important information might never be announced by a screen reader.
Agentic systems introduce another layer of complexity: An AI assistant may gather information, weigh options, make recommendations or carry out actions before presenting the outcome to the user. That places greater importance on transparency.
Users need clear visibility into what is happening, why particular decisions have been made and what information has informed those decisions.
This matters for everyone, with particular significance for disabled people who may already have experience of technology behaving in unexpected ways. Trust grows when users understand a system well enough to challenge it, correct it and override it when necessary.
Accessibility and autonomy
As AI agents become more capable, conversations about accessibility need to include autonomy, transparency and accountability alongside traditional interface design.
A blind user should be able to understand what an agent has done. Someone using voice interaction should be able to review and amend actions before they are completed. A person with cognitive impairments should be able to receive explanations in ways that support informed decision-making.
Users should have opportunities to pause actions, inspect recommendations and recover from mistakes. Together, these capabilities provide the control and clarity that meaningful accessibility requires when technology takes actions on a person’s behalf.
Designing with disabled people
In an ever-changing digital world, one lesson has remained consistent throughout my career in accessibility: the best outcomes emerge when disabled people are involved from the beginning. Agentic AI should follow the same principle.
Design teams, policymakers, regulators and technology providers all have important decisions to make about how these systems behave. Disabled people bring practical insight into independence, trust, usability and risk that cannot be replicated through technical analysis alone.
Their experiences should inform product design, testing, governance and evaluation. That involvement becomes particularly important because agentic AI will increasingly influence access to essential services, employment, education and financial products.
The next chapter of accessibility
Agentic AI has the potential to remove some of the friction that disabled people encounter every day. That possibility is genuinely exciting.
Whether it delivers on that promise will depend on decisions being made right now by technology providers, policymakers and organisations adopting these tools. Accessibility has a vital role to play in those decisions, and the people who understand disability, assistive technology and inclusive design need a seat at the table from the outset.
Getting this right could help more people participate fully in digital life. Overlooking accessibility could allow many familiar barriers to reappear in a different form.
Disabled people should help shape what comes next. The opportunity is too significant to approach without their knowledge and lived experience.
How AbilityNet can help
Many organisations are still working out what AI adoption means for accessibility, inclusion and customer trust. AbilityNet helps organisations build accessibility into AI strategy, procurement, design, testing and governance.
Through accessibility audits, disabled user testing, training and consultancy, we help organisations ensure that emerging technologies work for the broadest possible range of people.
Find out more about AbilityNet’s accessibility services and how we can help your organisation build accessibility and inclusion into AI initiatives from the outset.
Talk to us about accessible AI integrations
TechShare Pro 2026
This year's theme for Europe's largest digital accessibility is AI and Accessibility: Risk and Reward.
Where does your organisation stand on AI and accessibility?
Complete our annual Digital Accessibility Survey and help benchmark progress across the UK.
More from Robin Christopherson on AI and accessibility
Why accessibility is one of the most important AI procurement decisions you can make
- Published 24 Aug 2026. Inaccessible AI can create costly remediation, governance risks and poor customer experiences.
The hidden cost of AI bias: Why disability may be the ultimate test of trustworthy AI
- Published 17 Aug 2026. As AI adoption grows, disability is emerging as a critical test of trustworthy AI.
The EU AI Act has a message for accessibility teams: get involved earlier
- Published 10 Aug 2026. The EU AI Act's requirement for human oversight highlights the need to involve accessibility experts and disabled users much earlier in AI development.
Using AI tools to drive an accessible dev process
- Published 05 Aug 2026. How do we move beyond AI-generated code and start using AI to reduce accessibility debt before it reaches production?
AI could make financial services more inclusive. But only if accessibility is part of the infrastructure
- Published 27 July 2026. Why accessibility must be treated as core infrastructure if AI is to improve financial inclusion.
The trust gap: why AI's edge cases matter to customer loyalty
- Published 30 June 2026. Why accessibility and edge cases are critical to building trust in AI-powered services.
The delicious irony at the heart of trusting AI
- Published 17 June 2026. A personal reflection on trust, accessibility and the unexpected ways AI can both help and hinder.
Executive briefings for senior leaders
Robin offers invite-only executive briefings exploring AI, accessibility, governance, customer trust, regulation and digital inclusion.
