EU law must ensure trustworthy and accessible Artificial Intelligence (AI) for persons with disabilities!

EU law must ensure trustworthy and accessible Artificial Intelligence (AI) for persons with disabilities!

We call on the EU co-legislators to ensure trustworthy and accessible AI systems and practices that respect the right of persons with disabilities to non-discrimination, equality, privacy and data protection, and promote their social participation and independent living.

On the 21st of April, the European Commission (EC) released a proposal for new legislation establishing a legal framework on Artificial Intelligence (AI) in the EU. This proposal is a step in the right direction. However, much work still needs to be done to improve the existing regulation and establish trustworthiness of AI for persons with disabilities.

The EU and all member states, as signatories to the United Nations Convention on the Rights of Persons with Disabilities (CRPD), are legally obliged to protect persons with disabilities from discrimination, promote their equality, ensure equal access to information and communications technologies and systems, ensure respect for privacy, and protect their rights to education and health as well as work and employment. The existing proposal falls short in these areas. Additionally, the regulation should promote development of AI that will benefit independent living and social participation of all persons with disabilities.

In terms of accessibility, the EC proposal lacks mandatory accessibility requirements for AI systems and practices. Accessibility should be required for all AI systems, irrespective of perceived or actual level of risk (‘high-risk’ or ‘non-high-risk’) because lack of accessibility can risk life and well-being of persons with disabilities. We are disappointed that the text falls short of the commitments under the CRPD and lacks consistency with EU accessibility legislation such as the European Accessibility Act.

The proposal also allows the use of AI for job recruitment purposes, biometric identification, determining education access, by law enforcement, and to assess whether a person might pose a ‘health risk”, among others, opening the door to a wide range of possibilities of AI discrimination against persons with disabilities. Furthermore, under the existing EU Data Protection Regulation (GDPR), not all persons with disabilities will be able to refuse consent for processing their data. The proposal does not resolve this, nor does it sufficiently establish how to object to data collection or who to contact in case of a data breach.

To mitigate potential risk by AI technologies and ensure that they bring benefits for persons with disabilities, we call on the EU legislators, namely the European Commission, European Parliament, and Council of the EU, to:

  • Ensure horizontal and mainstreamed accessibility requirements for AI systems, irrespective of level of risk.
  • Prohibit several practices listed under Annex III related to the use of AI biometric identification and categorisation of natural persons, determining individuals’ opportunities to access education, employment, access to and enjoyment of essential private services and public services and benefits; and use of AI by law enforcement and for certain use in migration, asylum, and border control management (see specific suggestions in the Position Paper).
  • Ensure that privacy and data protection of all persons with disabilities are ensured when their data is processed by AI systems, and measures are set for individuals to be informed and object when their data is being gathered and processed by AI systems.
  • Ensure protection of fundamental rights of individuals within the context of AI application, including measures to flag issues, file complaints to authorised bodies, and seek remedies in case of abuse that are accessible for persons with disabilities.
  • Ensure ex ante human rights impact assessments for high-risk AI systems before putting them into use, as well as accessibility checks as part of conformity assessments.
  • Ensure that AI providers and users whose outputs affect individuals outside of the European Union are subject to same requirements as those whose outputs affect persons within the Union.
  • Ensure EU and Member States, as obliged by article 4.3 of the CRPD, closely consult with and actively involve persons with disabilities, through their representative organizations in the development, implementation, and monitoring of European and national AI policies. Doing so will not only reduce potential risks of AI but also ensure promotion of AI that can strengthen rights of persons with disabilities to independent living, freedom of movement and equal socio-economic participation.



Mher Hakobyan, EDF Accessibility Officer
Twitter: @armhak2003