The invisible gatekeepers: AI, opportunity and who gets left behind

Robin Christopherson | 29 Sep 2026

Over the past few years, the accessibility conversation around AI has largely focused on usability. Can a disabled person use an AI-powered product? Can they interact with the chatbot, complete the journey, understand the output and achieve their goal? Those questions still matter, and as AI moves deeper into the systems behind our services, another question is becoming harder to ignore: what happens when AI starts influencing who gets the opportunity in the first place?

Some of the most consequential AI systems operate in the background, beyond the parts of a service that users can see or control. They may help determine who gets shortlisted for a job, who receives additional support, who is flagged as a risk, who is routed to a specialist adviser, or who is prioritised for a service. Together, they are becoming part of the infrastructure of decision-making.

As an accessibility expert, I am concerned about what happens when these systems are built without a deep understanding of disability and human difference. As a blind person, I am equally concerned about decisions being shaped by systems that may never fully understand why someone behaves differently from the pattern they have learned to expect.

From digital barriers to decision barriers

Accessibility has traditionally focused on removing obstacles. Can someone access the website, complete the form and navigate the app using a screen reader, voice control, switch access or magnification software? Those remain essential questions, and AI adds a further layer to the challenge.A professional presenting notes, while looking at a screen showing 'AI decisions'

A disabled candidate may complete an accessible job application and still be filtered out by an automated assessment process. A customer may use an online banking service successfully while algorithms influence their eligibility, risk categorisation or route to support. A vulnerable customer may reach a chatbot, then be misunderstood and directed away from the human help they need.

Accessibility barriers can appear at the interface level and within the decision-making processes that sit behind a service. Accessibility has long helped people reach opportunities. AI adds another dimension because automated systems increasingly participate in the assessments, rankings and recommendations that influence what happens next.

Disability does not always fit the pattern

AI systems learn from data, and the implications of that simple fact can be profound. Many models are trained on datasets that reflect common patterns of behaviour, communication and interaction. Disabled people frequently use services in ways that fall outside those patterns.

Someone with a speech impairment may communicate differently. A neurodivergent person may interact differently with a customer service system. A blind person may navigate technology in ways that conventional analytics fail to represent. A person with a fluctuating condition may show changing patterns of activity that an automated system struggles to interpret.

These differences say nothing about capability, trustworthiness or suitability. Poorly designed systems can still convert difference into disadvantage, particularly when AI is used in recruitment, insurance, financial services, healthcare assessment, benefits administration or education. In these settings, a misunderstanding can have significant real-world consequences.

The concern is that systems trained on majority patterns may struggle to interpret minority experiences accurately. When their outputs influence a high-stakes decision, an apparently small gap in the data or evaluation process can affect someone's work, finances, health or access to essential support.

The problem with invisible decisions

Accessibility has progressed partly because barriers can be observed and demonstrated. If a website cannot be used with a keyboard, people can identify the problem. If a screen reader cannot read critical content, the issue can be reproduced and fixed. Automated decision-making can be much harder for the affected person to see. Wooden blocks with an outline of a brain fitting together nicely, however a round wooden block, that doesn't fit displays 'AI'

Someone may never know why they were rejected for a job interview, classified as higher risk or placed lower in a queue. They may have little visibility of the data used, the role AI played or how the system reached its conclusion. Even where an organisation acts in good faith, the process can become difficult to explain.

This creates a challenge around transparency, accountability and trust. People are more likely to trust decisions when they can understand how those decisions were reached and have a meaningful opportunity to challenge them when something appears wrong. Without that visibility, AI can become a silent gatekeeper whose reasoning remains hidden while its conclusions affect employment, finances, healthcare, education or access to support.

Why this particularly matters for high-stakes, highly regulated sectors

Sectors such as financial services provide a useful example because AI is increasingly being used across customer journeys and operational processes. Banks and insurers are exploring AI to improve customer service, support fraud detection, streamline assessments and help customers reach relevant products or assistance more quickly. At the same time, these organisations are both highly regulated and delivering services that are of critical importance to their customers.

Used with care, AI could help organisations identify customers who may need additional support, personalise communication and reduce friction in complex processes. The risks become more serious when accessibility and inclusion receive attention late in the design or procurement process.

When algorithms influence credit, insurance, risk or customer support, organisations need evidence that disability-related differences are being interpreted appropriately. Customers also need accessible explanations, a clear route to ask questions and timely access to a person when automated handling is unsuitable.

This is why I increasingly see accessibility as a governance issue as well as a design discipline. Decisions that shape customer outcomes are often made long before someone reaches the visible interface, across procurement, data, model selection, testing, monitoring and escalation.

Accessibility and power

This topic changes the scope of the accessibility conversation. For years, much of our effort has focused on helping disabled people reach digital services and complete the journeys within them. That work remains essential and unfinished, while AI brings questions about influence and power more clearly into view.

Who gets seen, believed, prioritised and offered an opportunity? Who can understand a decision and challenge it when something goes wrong? These questions sit at the intersection of accessibility, inclusion, consumer protection and human rights. They affect disabled people and many other groups whose lives and experiences may sit outside the patterns represented in training and evaluation data.

What inclusive AI decision-making requires

There is no single technical measure that can resolve these risks. Organisations need a combination of thoughtful design, careful governance and clear accountability, supported by practical evidence about how systems perform for disabled people.

  • Are disabled people included in design, testing and ongoing evaluation?
  • Can people understand AI-influenced decisions through explanations that are clear and accessible?
  • Is there a meaningful route to human review, with enough authority to change the outcome?
  • Are outcomes monitored for patterns of unintended exclusion or disadvantage?
  • Do procurement and governance teams assess accessibility and inclusion risks before deployment?
  • Can individuals challenge decisions without facing an inaccessible or exhausting process?

These questions need to be asked before a system becomes embedded in everyday operations. Once AI is woven through a service, hidden assumptions and unintended consequences can be difficult to trace, particularly when responsibility is spread across internal teams and external suppliers.

AI should widen access to opportunity

The promise of AI is substantial. It can improve access to information, support independent living, reduce friction and create new opportunities for millions of people. I see those benefits firsthand as a blind person who heavily relies upon technology in my daily and working life, and I want disabled people to share fully in what AI can offer. A team of professionals in a meeting, with screens showing code and 'AGI' display

AI is also influencing the pathways that lead to opportunity. It is becoming part of the machinery that helps determine who receives support, who gets selected, who is trusted and who is offered help. Accessibility therefore has a growing role in ensuring that the systems influencing these outcomes are fair, transparent and accountable.

Disabled people should be able to understand decisions that affect them and seek meaningful review when something appears wrong. Organisations need to explain how important decisions are made, provide workable routes for challenge and include disabled people from the outset when systems are designed, procured, tested and monitored.

People often think about accessibility in terms of doors being opened. In an AI-enabled world, however, we also need to pay attention to the gatekeepers granting or denying access to those same doors. As these systems become more influential, they must recognise human diversity in all its forms and remain open to scrutiny when their decisions carry real consequences.

AI should widen access to opportunity and support fairer outcomes. Achieving that will require deliberate choices, strong evidence and the involvement of disabled people throughout the life of the system.

How AbilityNet can help

Many organisations are now asking similar questions. How should accessibility teams engage with AI initiatives? How can AI be used responsibly within accessibility programmes? What governance, testing and assurance approaches are most effective? And how do you ensure disabled customers and colleagues are not left behind as adoption accelerates?

At AbilityNet, we help organisations answer those questions through accessibility strategy, audits, user testing with disabled people, training, consultancy and AI-focused accessibility guidance.

Explore AbilityNet's accessible AI integration and consultancy services or speak to our experts about an AI feature, customer journey or deployment that you want to assess. 

Talk to us about accessible AI integrations 

 

More from Robin Christopherson on AI and accessibility

AI and Vulnerable customers: The hidden risk of helpful AI

  • Published 22 Sep 2026. AI has huge potential to improve accessibility and inclusion, but it also raises important questions about trust, autonomy and vulnerability. 

How to build accessible AI: Insights from industry leaders

  • Published 15 Sep 2026. Accessibility leaders from across industries share their biggest lessons on AI adoption, governance, inclusive design and the risks of scaling accessibility at speed.

Five questions to ask about every AI deployment

  • Published 8 Sep 2026. AI could generate inaccessible websites, apps and content at such scale that organisations end up with an even bigger accessibility backlog than before.

Accessible AI: Who is responsible when AI acts on our behalf?

  • Published 2 Sep 2026. As AI takes on more tasks, accessibility becomes more important than ever. 

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.