Report: AI Accessibility, Deaf Communities, and Shared Agreement

AuthorTim Scannell - Accessibility Consultant (Deaf & British Sign Language User)

1. Purpose of This Report

This report explores how AI accessibility systems affect Deaf people and explains why shared agreement between the Deaf community, AI providers, and organisations (including government, funders, and policy decision-makers) is essential before any AI system is used.

Its purpose is to support better understanding, collaboration, and safer outcomes by identifying common challenges, highlighting areas of risk, and setting out positive conditions for ethical and effective AI accessibility.

This report is offered in a constructive spirit, with the aim of supporting learning and improvement across all parties.

2. Professional Context

I am a Deaf professional and British Sign Language (BSL) user working in accessibility and inclusion. My work is based on lived experience, professional evaluation, and direct observation of how technology functions for Deaf people in real-world situations.

My role focuses on:

• Assessing whether accessibility solutions genuinely work for Deaf users

• Identifying risks, gaps, and unintended impacts

• Reviewing claims made about AI accessibility

• Supporting the inclusion of Deaf user perspectives in decision-making

I am not an AI engineer, developer, or vendor. I do not build or sell AI systems, and I do not replace interpreters or human communication support. My contribution is independent and user-focused.

3. The Three-Party Relationship

AI accessibility systems typically involve three groups:

• The Deaf Community – the people directly affected

• AI Providers – those designing, training, or supplying AI systems

• Organisations – government bodies, funders, investors, policy owners, and decision-makers

  
  

For accessibility to be meaningful and safe, these three groups need to work together on equal and respectful terms. When alignment is missing, even well-intentioned efforts may not achieve their intended outcomes.

4. Deaf Community Perspective

From lived experience, Deaf communities commonly value:

• Direct consultation in sign language, not text-only engagement

• Paid participation, recognising lived experience as professional expertise

• Human-led access, alongside appropriate use of technology

• Sign language-first approaches, rather than adaptations of hearing-centric systems

• Clear consent and transparency around data use

• Independent review and accountability

• Respect for language and culture, alongside innovation

  
  

When these elements are present, trust grows and accessibility improves.

5. AI Provider Considerations

Many AI providers are working to innovate quickly and respond to accessibility challenges. At the same time, lived experience shows that:

• Deaf users are sometimes involved later in the process

• Feedback may be collected indirectly rather than through direct engagement

• Claims of sign language support can benefit from clearer, independent validation

• Responsibilities around error handling are not always visible

• Automation may unintentionally replace human expertise

  
  

Addressing these areas creates opportunities for stronger, more trusted solutions.

6. Organisational and Government Role

Organisations and decision-makers play a vital role in shaping outcomes. Positive impact is strengthened when organisations:

• Involve Deaf communities early and meaningfully

• Pair innovation with clear safeguards

• Seek verification alongside compliance claims

• Actively uphold language and accessibility commitments

• Operate transparently and with shared accountability

• Include Deaf representation in governance and advisory roles

  
  

This approach helps ensure responsibility is shared rather than passed on to users.

7. The Shared Agreement Principle

Core Principle

AI accessibility systems work best when all three parties explicitly agree:

• Deaf Community ✔

• AI Provider ✔

• Organisation / Decision-Maker ✔

If any one party is not ready to agree, it signals a need to pause, review, and improve together.

Accessibility is not a majority vote — it is a shared responsibility.

Why Agreement Matters

• Community agreement supports legitimacy and trust

• Verified AI supports safety and reliability

• Organisational accountability supports responsibility

If one part is missing, the system needs more work. This is not a failure — it is an opportunity to strengthen outcomes.

8. High-Risk Areas

Based on lived experience, additional care is needed when considering AI in:

• Legal and justice contexts

• Healthcare and medical settings

• Education and assessment

In these areas, maintaining human-led access is especially important due to the potential impact of errors.

9. Impact of Moving Forward Without Agreement

When AI is introduced without full alignment:

• Miscommunication can increase

• Human access may be reduced unintentionally

• Cultural and linguistic needs may be overlooked

• Legal and ethical risks may rise

• Trust can be weakened

These outcomes are avoidable through collaboration and shared decision-making.

10. Positive Conditions for Use

Before considering AI accessibility systems, the following conditions support success:

• Deaf communities are engaged early, in sign language

• Deaf professionals are hired and paid for their expertise

• AI systems are independently reviewed and verified

• Human access remains protected

• Organisations accept shared accountability

• All three parties agree the system is ready

  
  

These conditions support trust, safety, and long-term sustainability.

11. Final Statement

AI can support accessibility when used thoughtfully and responsibly. It should not judge, diagnose, teach, or decide for Deaf people.

True accessibility is built on human responsibility, community consent, and shared agreement.When any party raises concerns, that moment should be seen not as a barrier, but as an invitation to work together and improve.