AI and disabled workers: barriers, accommodations and the law

AI tools at work cut both ways for disabled people. Speech-to-text, screen readers, captioning and writing assistants can remove barriers that used to make a job hard to do. The same wave of software can also add new barriers: an online test that a screen reader cannot use, a video interview scored on facial expressions or speech, or a productivity score that counts a needed break as idle time. This page sets out what the law says about both sides, for workers and for employers.

The short answer: in the US, the Americans with Disabilities Act applies to hiring technology. Employers must not use tools that unfairly screen out qualified disabled people, must make sure tests measure job skills rather than a disability, and must provide reasonable accommodations, including during hiring, unless that would be an undue hardship. In the EU, the AI Act bans AI that infers workers' emotions, except for medical or safety reasons, and requires providers of high-risk AI, which includes many employment uses, to meet accessibility requirements.

How AI tools can screen people out

The US Department of Justice guidance on algorithms, AI and disability discrimination in hiring, published on 12 May 2022 and still on ADA.gov, gives concrete examples (U.S. Department of Justice, ADA.gov):

• Learning from the current workforce: tools that predict good employees by comparing applicants with current staff can discriminate, because disabled people have historically been excluded from many jobs and may not be part of that staff.
• Facial and voice analysis: people with disabilities such as autism or speech impairments may be screened out even if they are qualified.
• Inaccessible tests: an applicant with a vision impairment may be passed over because they do poorly on a computer-based test that requires them to see, even though they can do the job.
• Incompatible software: an online interview program that does not work with a blind applicant's screen reader.

The guidance notes that a disability affects each person differently, so a tool checked for one type of disability can still discriminate against people with another. It also says employers are responsible when they use another company's discriminatory hiring technology.

What the ADA requires

According to the same guidance, the ADA applies to all parts of employment, including how an employer selects, tests or promotes employees (ADA.gov). The DOJ enforces it for state and local government employers and the EEOC for private and federal employers. In summary:

• Screening out: employers should examine hiring technologies before use, and regularly when in use, to check whether they screen out disabled people who can do the essential functions of the job with or without reasonable accommodation.
• Tests measure skills, not disability: tests, games and personality assessments must measure the relevant skills and abilities rather than impaired sensory, manual or speaking skills they do not seek to measure.
• Accommodations: reasonable accommodations must be provided, including during hiring, unless they would be an undue hardship.
• Medical questions: hiring technologies must not unlawfully seek medical or disability-related information or amount to a medical exam.

The guidance suggests telling applicants what technology is used and how they will be evaluated, giving enough information for them to decide whether to ask for an accommodation, and having clear procedures for asking, so that asking does not hurt their chances. It also explains that people who believe they faced disability discrimination at work can file a charge of discrimination with the EEOC. See AI in hiring.

Beyond hiring: monitoring and performance scores

The same ADA principles apply once someone is in the job, because the law covers all parts of employment. A monitoring or scoring system that treats an accommodated schedule, extra breaks or assistive software as low productivity raises the same question as a hiring test: is it measuring the work, or the disability? Managers should know who to call when a system flag involves an accommodation, and should not act on a score without checking. See AI monitoring at work and AI in performance reviews.

In the EU

The AI Act prohibits the placing on the market, putting into service or use of AI systems to infer emotions of a person in the areas of workplace and education institutions, except where intended for medical or safety reasons (Regulation (EU) 2024/1689, Article 5(1)(f)). It separately prohibits AI that exploits vulnerabilities due to age, disability or a specific social or economic situation in a way that materially distorts behaviour and causes or is reasonably likely to cause significant harm (Article 5(1)(b)). These prohibitions have applied since 2 February 2025.

AI systems used in recruitment, promotion, termination, task allocation and performance monitoring are listed as high-risk (EU AI Act, Annex III, point 4), and providers of high-risk systems must ensure they comply with EU accessibility requirements under Directives (EU) 2016/2102 and 2019/882 (Article 16(l)). The high-risk obligations for these employment uses apply from 2 December 2027 (European Commission). In Great Britain, see AI at work and the Equality Act 2010 for the duty to make reasonable adjustments, and what counts as a disability under the Act. See AI at work and workers' rights.

Practical steps

• Workers: ask what technology will assess you and how; if a tool does not work for you, ask for an accommodation in writing and keep a copy. You can also ask what data a system holds on you; see how to request the data AI systems hold on you.
• Employers: test new tools with assistive technology before buying, ask vendors how they checked for effects on different disabilities, offer an alternative route for any automated assessment, and make the accommodation process visible at every step.

In Great Britain, see reasonable adjustments for AI tools at work. See also AI and older workers. This page is general information, not legal advice.