Rapporteur:

Dovilė JUODKAITĖ

Advisor	Álvaro Antonio COUCEIRO FARJAS (to the rapporteur, Group III)
Request from the Polish Presidency of the Council of the EU	Letter of 6.9.2024
Legal basis	Article 304 of the Treaty on the Functioning of the European Union
Section responsible	Employment, Social Affairs and Citizenship
Adopted in section	11.3.2025
Adopted at plenary session	26.3.2025
Plenary session No	595
Outcome of vote (for/against/abstentions)	150/0/2

1.   Conclusions and recommendations

1.1.

The EESC believes it is crucial to include persons with disabilities throughout the process of designing, developing, assessing and implementing new technologies and AI-based applications. This involvement is key to mitigating potential discriminatory risks.

1.2.

The EESC invites the Member States and the EU to audit and evaluate to what extent AI applications are inclusive in critical fields. Positive practices that improve the inclusiveness of AI-based applications to the benefit of groups that might be victims of discrimination must be shared.

1.3.

The EESC recognises the advances brought by the EU legal framework on the protection of persons with disabilities from unacceptable and high risk AI systems, protection of sensitive personal information, and the prohibition of manipulative AI programmes. Nevertheless, it believes further follow-up is required to analyse the possible need to protect persons with disabilities from the harmful application of supposedly limited risk systems, particularly in the field of employment and recruitment.

1.4.

The EESC underlines the importance of ensuring the general accessibility of services and goods that apply new technologies and AI-based solutions in accordance with Directive (EU) 2019/882 of the European Parliament and of the Council (1), Directive (EU) 2016/2102 of the European Parliament and of the Council (2) and related accessibility standards.

1.5.

The EESC believes that Member States must ensure that the opportunities generated by new technological developments and AI applications are translated by employers into accessibility improvements and reasonable accommodation for employees with disabilities, and that the European Commission must promote this. While supporting the transition to new digital tools, current positive assistive technologies should not be discarded with the new alternatives.

1.6.

The EESC recommends increasing public expenditure on research and development of AI-based solutions for accessibility and technological assistance, and improving public knowledge about these technological advances, as well as existing government support for acquiring that knowledge and training for all users.

1.7.

The EESC urges Member States to implement and increase the existing programmes on digital skilling for persons with disabilities, to improve their current situation and qualifications in these fields, while working to increase the inclusion of persons with disabilities in tertiary education and vocational training in technological sectors. The Commission and the EU Member States must collect EU-wide data analysing these rates and the disability gap.

1.8.

The EESC highlights the need to promote reasonable accommodation for employees with disabilities in the workplace, including in public and private recruitment processes, skilling and vocational training as well as in tertiary education, given the current obstacles that lead to the extremely low tertiary education rates among persons with disabilities. Recruitment applications based on new technologies and AI must avoid potential discriminatory situations and be accessible, or include accessible alternatives.

2.   Background

2.1.

Ahead of its presidency of the Council of the EU, on 6 September 2024, the Minister for the European Union of the Republic of Poland requested a number of exploratory opinions from the European Economic and Social Committee (EESC), including an opinion on the Inclusion of persons with disabilities in the context of the development of new technologies and AI – possibilities, challenges, risks and opportunities.

2.2.

This exploratory opinion summarises topics related to the current situation and the potential impact of technological advances and artificial intelligence (AI) with regard to the employment of persons with disabilities in the EU. It responds to the questions posed by the presidency relating to the risks of discrimination against people with disabilities posed by the use of AI algorithms and other new technologies in the labour market.

3.   General comments

3.1.

Article 27 of the United Nations Convention on the Rights of Persons with Disabilities (UNCRPD), which was ratified both by the EU in 2011 (3) and by each Member State, recognises the right of persons with disabilities to work on an equal basis with others. It sets out the requirements for achieving this, which include ensuring access to training, and providing reasonable accommodation in the workplace, and removing any barriers that may prevent persons with disabilities from participating fully, including barriers in information and communications technologies and systems (4).

3.2.

New technologies and AI allow for new opportunities for persons with disabilities when it comes to education, employment and participation in society. However, the solutions developed have to be accessible and persons with disabilities considered when developing them. In this regard, the Accessibility Act (5) sets the basis for the minimum accessibility requirements, reiterating the need to increase accessibility of available and future technologies, including AI. The AI Act (6) acknowledges this requirement as a priority, considering it essential that providers comply with accessibility requirements fully. The development of specific accessibility standards for AI technologies should be explored (7).

3.3.

In 2022, the European Commission launched the Disability Employment Package so as to advance in the implementation of measures aimed at complying with the UNCRPD in the EU, in line with its Strategy for the Rights of Persons with Disabilities 2021-2030 (8).

3.4.

However, at EU level, only 50,8 % of persons with disabilities are employed, compared to 74,8 % of persons without disabilities, and just 20 % of women and 29 % of men with disabilities are working full time (9). When working, persons with disabilities in the EU earn, on average, 17,76 % less annually than persons without disabilities, equivalent to 3 200 PPS (10). It is thus clear that the EU and its Member States have to improve the inclusion of persons with disabilities in the labour market, focusing on public support, education and training, accessibility and reasonable accommodation and eradication of discriminatory attitudes.

3.5.

Low levels of digital skills and the digitalisation of the labour market, together with the substitution of low-value jobs by AI solutions risks increasing the socio-economic exclusion of persons with disabilities if the digital gap is not closed.

3.5.1.

The digitalisation process and new technologies in general facilitate the resolution of employability problems and improve working processes but, at the same time, open the door to new challenges and potential risks if they do not take into account the situation of every potential person affected by them.

3.5.2.

First, there is a significant gap regarding the use of these technologies in the labour market. While around 92 % of positions in the current job market require digital skills, and the EU and its Member States have stepped up their efforts to increase general digitalisation, persons with disabilities tend to have a lower level of digital literacy and training, as well as more limited access to internet connections and technological tools (11). Digitalisation processes must be inclusive to avoid the exclusion of vulnerable groups, as stated in the previous opinion (12).

3.5.3.

Currently in the EU only 18 % of persons with disabilities complete tertiary education, compared to 39 % of the general population (13). Together with the lower digital skills rates, this creates difficult prospects for their future labour and social inclusion, all the more so since digital skills also foster educational inclusion, and participation in cultural, leisure and social activities (14).

3.5.4.

Labour market tendencies show that lower-skilled jobs are being substituted by new technologies, and some persons working in these fields are already losing their jobs. Globally, 2,3 % of jobs are at risk of extinction due to automated technologies (15). This tendency particularly affects women and persons with disabilities, as they tend to work in lower-value sectors due to the aforementioned skills and digital gaps. The ongoing digitalisation processes in the fields of education and training risk exacerbating their current already low inclusion levels, and may diminish their subsequent potential employability and social inclusion possibilities.

3.5.5.

Therefore, it is crucial for the EU and its Member States to increase funding for the improvement of digital skills of persons with disabilities, ensuring that they can access the tools needed. Educational tools on digital skills based on AI, such as online learning platforms and adaptative learning technology, can be used as accessible and affordable solutions for the education of persons with disabilities who are distanced from mainstream education, but they should not hinder efforts to promote inclusive education.

3.5.6.

In this regard, Member States must increase their investment in improving the inclusion of persons with disabilities in education and training. The current low educational and training rates, at all levels, are the main obstacle hindering labour inclusion later on.

3.6.

AI offers opportunities for improving accessibility and technological assistance, but, while supporting the transition to new user-friendly digital tools, these solutions must not forcedly replace existing accepted assistive technologies, and interoperability between them has to be ensured.

3.6.1.

AI opens new doors to the simplification of labour-related procedures, as well as the structure of current jobs, if they are developed with a ‘Design-for-All’ approach. These advances, together with the development of AI-based accessibility adaptations, may facilitate the inclusion of persons with disabilities in the labour market, as well as independent living. AI programmes enable assistive technologies to adapt to different scenarios and disabilities, whereas pre-existing assistive devices tend to be single-purposed (16). As examples, voice-to-text technology can help a writer with limited mobility, and AI-supported screen readers can improve the formatting of documents written by persons who are visually impaired.

3.6.2.

Developers of AI-based technology should ensure effective participation and consultation of persons with disabilities throughout the process with the purpose of improving accessibility. Similarly, the deployment of AI-based technology at a workplace should be facilitated by involving workers with disabilities through general and fair social dialogue. The exchange of good practices of AI-based inclusion solutions amongst EU Member States must as well be enforced.

3.6.3.

At the same time, while supporting the transition to new digital tools can be positive, the potential implementation of new technologies and AI in order to improve accessibility and reasonable accommodation does not mean that companies and employers should impose them as replacements for existing functioning technologies which are commonly used and accepted by persons with disabilities. Also, interoperability between new AI-powered solutions and existing hardware and assistive technology devices – such as AI-supported screen readers, or manual audio-to-text transcriptions with AI assistance – should be considered, to ensure that they can be adopted by persons with disabilities (17).

3.7.

AI-based recruitment and employment-assessment applications can be discriminatory if they do not factor in the particular issues and common educational and labour-related obstacles that persons with disabilities face. The implementation of reasonable accommodation for persons with disabilities must be stepped up.

3.7.1.

While providing solutions to current problems, AI also poses risks and challenges. Understanding this potential harm, the AI Act recognises the risk of discrimination against persons with disabilities. Biometric identification may lead to biased results, and AI systems can exploit vulnerabilities and impair autonomy. It also points to the risks when AI is used in recruitment processes and performance monitoring, as this may perpetuate discrimination and undermine fundamental rights to data protection and privacy. Such uses of new technologies are classified as posing a high risk to the rights and freedom of data subjects.

3.7.2.

A specific risk may also emerge whenever AI tools are used to analyse CV databases for more agile recruitment processes, whenever these tools do not factor in the particular issues that persons with disabilities face within the education system and in their careers. Within an automated system, habilitation and rehabilitation career gaps, longer training and education periods for disability-related reasons or participation in specialised education settings might count against people. A human-based analysis or oversight of AI tools with a disability perspective must back up such systems in order to avoid potential discrimination.

3.7.3.

Other risks lie in the possible implementation of digital recruitment processes through non-accessible tools, not taking account of reasonable accommodation needs. The forced implementation of new technology-based recruitment processes may also hinder the participation of candidates with disabilities, if there are no accessible alternatives.

3.7.4.

Overall, reasonable accommodation in the workplace is not commonly applied within European companies, due to misunderstanding regarding its meaning and compulsory application by employers, its costs and complexity or the available public support. Up to 70 % of European employers do not have human resource guidelines on reasonable accommodation, including in recruitment processes, and 76 % are unaware of public policies or programmes supporting its implementation in the workplace (18). Hence, improvement of reasonable accommodation lies not just in the development of new public policies, but in the correct dissemination and implementation of those already in place.

3.7.5.

The European Commission recently published recommendations and guidelines on reasonable accommodation at work (19), as part of the Disability Employment Package. These principles should guide public authorities as employers, and be disseminated by the Member States. Examples within the document of good practices include the provision of assistive technologies and other new technology-based solutions that can be replicated by other employers.

3.8.

Persons with disabilities must be included and the disability perspective considered in the development of AI algorithms and new technologies.

3.8.1.

It is crucial for the EU, its Member States and the social partners to understand better how the application of new technologies and AI procedures within the labour market require a disability perspective. Understanding the implications for persons with disabilities is essential from the very first design of AI algorithms, and must continue in the further development of solutions by companies and the assessment of their applications.

3.8.2.

In order to apply the disability perspective throughout the process of AI system development and evaluation, digital developers must always consult entities representing persons with disabilities, hiring persons with disabilities for specific assessment and consultation tasks. The Commission and the Member States must ensure this process and promote public programmes aimed at improving the inclusion of persons with disabilities in these fields.

3.8.3.

Member States must also ensure that universities include accessibility and inclusive design training within computer science curricula, as well as in other engineering courses, supporting this with national and EU funds, given that it would improve the relevance of specialised solutions and the accessibility of mainstream AI from the early stages of development.

3.9.

AI poses specific risks of discrimination for persons with disabilities in the banking and insurance sectors.

3.9.1.

Banking and insurance are two sectors where the application of different AI models for the analysis of persons involves clear discrimination risks, given the irregular income patterns or employment gaps, amongst other things. AI is used to analyse personal information to apply specific services, prices or coverage. The use of automatic tools to approve or deny services, as well as to simplify, expedite or refine products may therefore be risky from the perspective of discrimination or personal data protection.

3.9.2.

The AI Act highlights that AI systems designed for risk assessment and pricing for health and life insurance are considered high risk, as they may infringe the fundamental rights of particular persons and lead to serious consequences for people’s life and health, including financial exclusion and discrimination, if they are not duly designed, developed and used. Hence, it upholds that banking and insurance private entities deploying AI-based technologies that fall into the high-risk category must carry out fundamental rights impact assessments prior to their application.

3.9.3.

Member States must step up public programmes and incentives to increase financial access for persons with disabilities in the EU, both in terms of physical, informative and communicative accessibility and in the prevention of discrimination. Finance training for persons with disabilities and staff of financial institutions on disability awareness may also help reduce barriers (20).

---

(1)  Directive (EU) 2019/882 of the European Parliament and of the Council of 17 April 2019 on the accessibility requirements for products and services (OJ L 151, 7.6.2019, p. 70, ELI: http://data.europa.eu/eli/dir/2019/882/oj).

(2)  Directive (EU) 2016/2102 of the European Parliament and of the Council of 26 October 2016 on the accessibility of the websites and mobile applications of public sector bodies (OJ L 327, 2.12.2016, p. 1, ELI: http://data.europa.eu/eli/dir/2016/2102/oj).

(3)  European Commission Vice-President Reding, Press Release, EU ratifies UN convention on disability rights, 5 January 2011, http://europa.eu/rapid/pressReleasesAction.do?reference=IP/11/4.

(4)  United Nations. (2006). ‘Convention on the Rights of Persons with Disabilities’. Treaty Series, 2515, 3.

(5)  Directive (EU) 2019/882 of the European Parliament and of the Council of 17 April 2019 on the accessibility requirements for products and services (OJ L 151, 7.6.2019, p. 70, ELI: http://data.europa.eu/eli/dir/2019/882/oj).

(6)  Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act) (OJ L, 2024/1689, 12.7.2024, ELI: http://data.europa.eu/eli/reg/2024/1689/oj).

(7)  The Technical Guide on Accessible and Equitable Artificial Intelligence Systems from the Government of Canada could serve as a basis for the EU: https://accessible.canada.ca/centre-of-expertise/information-and-communication-technologies/technical-guide-accessible-artificial-intelligence-systems#s4.

(8)  European Commission (2021). Union of Equality Strategy for the Rights of Persons with Disabilities 2021-2030.

(9)  Hammersley, H. & Buchanan, J. (2023). The Right to Work: The Employment Situation of Persons with Disabilities in Europe, European Disability Forum. Pp 35. https://www.edf-feph.org/content/uploads/2023/05/hr7_2023_press-accessible.pdf.

(10)  Purchasing power standard (PPS) is an artificial currency that accounts for price differences between Member States. European Commission (2021). Union of Equality Strategy for the Rights of Persons with Disabilities 2021-2030.

(11)  Lulli, R., Sinicato, F., Couceiro, A. (2024). Digital Skills, Accommodation and Technological Assistance for Employment: Supporting the inclusion of persons with disabilities in the open labour market. Pp. 26-27.

(12)   OJ C 374, 16.9.2021, p. 11.

(13)  Eurostat. (2022). How many citizens had basic digital skills in 2021?, https://ec.europa.eu/eurostat/web/products- eurostat-news/-/ddn-20220330-1.

(14)  UNESCO (2014). Role of ICTs for Persons with Disabilities, International Conference, 24-26 November 2014, New Delhi, India, https://unesdoc.unesco.org/ark:/48223/pf0000229818.

(15)  EESC and ILO (2025). Conference on Social justice in the digital era: AI’s impact on work and society – A contribution to the Global Coalition for Social Justice. Brussels, Belgium, https://www.eesc.europa.eu/en/agenda/our-events/events/social-justice-digital-era-ais-impact-work-and-society-contribution-global-coalition-social-justice.

(16)  Touzet, C. (2023), ‘Using AI to support people with disability in the labour market: Opportunities and challenges’, OECD Artificial Intelligence Papers, No. 7, OECD Publishing, Paris, https://doi.org/10.1787/008b32b7-en.

(17)   OJ C 374, 16.9.2021, p. 11.

(18)  European Commission (2021). Union of Equality Strategy for the Rights of Persons with Disabilities 2021-2030, p. 26-27.

(19)  European Commission: Directorate-General for Employment, Social Affairs and Inclusion. (2024). Reasonable accommodation at work: guidelines and good practices. Publications Office of the European Union, https://data.europa.eu/doi/10.2767/359088.

(20)  AFI (2023), Financial Inclusion for Persons With Disabilities. Insights from AFI Members’ Strategies and Policies. Special Report, https://www.afi-global.org/publication/financial-inclusion-for-persons-with-disabilities/.