AI Job-Matching for Adults With Disabilities: Tool or Gatekeeper?

Listen to this article

Played by your own device. Nothing is recorded and nothing is sent anywhere.

Picture a hiring chatbot that screens applicants before a human ever sees them. It scans for gaps in employment history and flags the wide ones. To the system, a two-year gap is a red flag, a sign of someone who stopped working. It does not know, and is not built to ask, that the gap was medical leave, a hospital stay, or the time it took to recover and find an accommodating employer. It just sees the gap and screens the person out.

That is not a hypothetical. The United States Equal Employment Opportunity Commission used almost exactly that example in technical assistance it issued on 12 May 2022, warning that AI hiring tools can illegally discriminate against people with disabilities. That document is no longer on the EEOC website, so the wording below is quoted from an archived copy captured on 26 January 2025. The regulator’s own example was a chatbot that “rejects all applicants who, during the course of their ‘conversation’ with the chatbot, indicate that they have significant gaps in their employment history.” Where the gap was caused by a disability, the EEOC wrote, “the chatbot may function to screen out that person because of the disability”. The Department of Justice published companion guidance the same day.

Two things about that guidance are worth knowing before leaning on it. The two agencies cover different employers, and the Justice Department sets the split out on its own page: it “enforces disability discrimination laws with respect to state and local government employers,” while the Equal Employment Opportunity Commission “enforces disability discrimination laws with respect to employers in the private sector and the federal government.” And both documents are guidance rather than law. The Justice Department states on its own page that its guidance documents “have no legally binding effect” and “may be rescinded or modified” at the department’s discretion. What binds is the statute underneath. The two documents are no longer in the same place. The Justice Department guidance was still published at ada.gov when this article was checked on 24 August 2026. The EEOC document was not. The EEOC’s own link to it returns a page-not-found error, as does the EEOC’s Artificial Intelligence and Algorithmic Fairness Initiative page. Archived captures show the EEOC page live on 25 January 2025 and gone by 27 January 2025. The EEOC has not published a notice rescinding the document, and taking a technical assistance document off a website does not change the Americans with Disabilities Act, which is the law the document was explaining. Treat the EEOC material below as a record of what the agency said in 2022, not as guidance the agency currently publishes.

How job-matching AI can screen you out

Employers, in the Justice Department’s words, “increasingly use hiring technologies to help them select new employees”. Resume-screening algorithms rank applicants. Chatbots conduct first-round interviews. Video tools analyze how candidates speak and move. Game-based assessments score cognitive and personality traits. The pitch is efficiency: sort thousands of applicants fast and surface the best ones.

The problem is what these systems learn to treat as “best.” They are trained on data about who has been hired and who has succeeded in the past. If the past favoured candidates with unbroken work histories, fast verbal responses, particular facial expressions, or a standard pace through an assessment, the AI learns to favour the same things. People with disabilities are disproportionately the ones who do not fit that mould, and not because they cannot do the job.

The Justice Department names the mechanism directly. Some hiring technologies “try to predict who will be a good employee by comparing applicants to current successful employees,” and “because people with disabilities have historically been excluded from many jobs and may not be a part of the employer’s current staff, this may result in discrimination.” Its worked examples are concrete: facial and voice analysis used to evaluate applicants may screen out “people with disabilities like autism or speech impairments,” and an applicant with a vision impairment “may get passed over for a staff assistant job because they do poorly on a computer-based test that requires them to see, even though that applicant is able to do the job.” The governing principle it sets out is that a test must measure “only the relevant skills and abilities of an applicant, rather than reflecting the applicant’s impaired sensory, manual, or speaking skills that the tests do not seek to measure.” In each case the candidate could do the job, with or without accommodation, and gets filtered out before anyone finds out.

The Canadian frame

The detailed regulator guidance on this comes from the United States, but the legal principle is just as firm in Canada. Discrimination in hiring on the basis of disability is prohibited under the Canadian Human Rights Act federally and under provincial human rights codes across the country. The duty to accommodate reaches the hiring process itself, not only the job once you have it. The Canadian Human Rights Commission gives a job-application example of its own: “asking all job applicants to pass a written test may not be fair to a person with a visual disability,” and in such cases “the duty to accommodate may require that alternate arrangements be made to ensure that a person or group can fully participate.” An employer cannot escape that duty by outsourcing the screening to software. If an AI tool screens you out in a way that a human process would have been required to accommodate, the legal obligation does not disappear because a machine made the call.

The duty has a limit, and it is worth knowing rather than being surprised by. The Commission calls it undue hardship: accommodation may not be required where it would cost too much or create health or safety risks, and an employer claiming undue hardship has to provide sufficient evidence. The Commission also notes that a person cannot hold out for the ideal accommodation if a reasonable one is offered.

Ontario has gone further than the rest of the country on the visibility problem. Since 1 January 2026, under the Employment Standards Act, 2000, an Ontario employer advertising a publicly advertised job posting must include “a statement disclosing the employer’s use, if any, of artificial intelligence to screen, assess or select applicants for the position.” The rule applies to employers with 25 or more employees on the day the posting goes up, and not to the Crown, Crown agencies, or bodies whose members are all appointed by the Crown. It applies just as much when the employer hires a recruiting firm to do the screening; the employer stays responsible for the posting.

Know what that disclosure does and does not give you. Ontario is explicit that an employer need not describe the system: “It is not necessary to provide a detailed description of the artificial intelligence system or the employer’s use of the system. It is enough for the employer to state that artificial intelligence is used to screen, assess or select applicants.” So the posting can tell you AI is in the process without telling you what it measures. One further Ontario rule is useful on its own: if you are interviewed for a publicly advertised job posting, the employer must tell you within 45 days of the interview, or of the last interview, whether a hiring decision has been made.

Outside Ontario the general protections still apply, but you often cannot see the tool that rejected you, which makes the rights hard to use unless you know to ask.

What to know before you apply

This is general information, not legal advice; for a specific situation, consult a lawyer or your human rights commission. You cannot control an employer’s hiring software, but you can control how you go in. A few things are worth knowing before you click apply.

  • Employment gaps are not a confession. Human rights law across Canada restricts what an employer may ask about disability before a job is offered, and how much detail you volunteer about a gap is your decision. What an employer may lawfully ask in a particular province is a question for a lawyer or that province’s human rights commission. A brief, neutral line (“focused on health,” “personal leave”) is enough if you choose to address it at all.
  • Canadian human rights law generally allows a person to request accommodation in the application process, not just the job. If an assessment, video interview, or timed test would disadvantage someone because of disability, they can ask for an alternative format, and that request triggers the duty to accommodate.
  • Accommodation in hiring is a right, not a favour. Framing matters. You are not asking the employer to do you a kindness. You are asking them to meet an obligation they already have under human rights law.
  • Look for the human contact. Application portals that route everything through a bot with no named recruiter make accommodation requests harder. A posting with a real contact person is a better sign.

What to ask employers about their AI

You are allowed to ask how a company evaluates you. Thoughtful employers will have answers. Evasive ones tell you something too. The Justice Department guidance sets out what employers using these tools may need to do so that applicants can actually get accommodations. It is United States guidance and does not bind a Canadian employer, but it is a plain description of what good practice looks like and a reasonable thing to ask about: “telling applicants about the type of technology being used and how the applicants will be evaluated,” “providing enough information to applicants so that they may decide whether to seek a reasonable accommodation,” and “providing and implementing clear procedures for requesting reasonable accommodations and making sure that asking for one does not hurt the applicant’s chance of getting the job.” Worth asking, especially once you are far enough in the process to have a contact:

  • Do you use automated or AI tools to screen applications or conduct assessments? A straight question deserves a straight answer.
  • If a timed assessment or video interview is part of this, is an alternative format available as an accommodation? Ask before you take it, not after.
  • Is there a point where a person reviews applications, or does the software decide who advances?
  • Who do I contact to request an accommodation in the hiring process? Get a name and put the request in writing.

If you believe an automated tool screened you out because of disability, the route is the official one. In Canada, you can bring a complaint to the Canadian Human Rights Commission for federally regulated employers, or to your provincial human rights tribunal for most others. Proof that the employer meant to discriminate is not what these processes turn on. Setting out how discrimination is established under the Ontario Human Rights Code, the Ontario Human Rights Commission states it plainly: “Discrimination does not have to be intentional. Intent is irrelevant for establishing that discrimination occurred.” The Commission adds that a protected ground “need only be one factor, of possibly several,” in the treatment for discrimination to be found. That is the Ontario Commission describing the Ontario Code. The tribunal or commission in your own province sets out its own test, and the Canadian Human Rights Commission does so for federally regulated employers.

A missing AI disclosure is a different complaint from a discrimination complaint, and it goes somewhere else. In Ontario, potential non-compliance with the job posting requirements is reported to the Ministry of Labour, Immigration, Training and Skills Development at PubliclyAdvertisedJP@ontario.ca, which asks for the employer name and contact information, the physical address of the business, and a link to or copy of the posting.

The takeaway

AI job-matching can be a useful tool or a gate that closes before you reach it, and which one it is depends entirely on how the employer built and deployed it. You cannot audit their software, but you are not powerless either. Know that a gap is not a verdict, that you can request accommodation in the process itself, and that the duty to accommodate follows the job application straight through any algorithm the employer chooses to use. Ask whether AI is screening you. Ask for an alternative if you need one. And if a machine shuts a qualified candidate out on disability grounds, that is the kind of decision human rights law was built to undo.

Related reading: coding with a disability: the tools, the workflows, the real constraints.

Sources

Similar Posts