Does your hiring AI comply with accessibility law, not just privacy law?

An employer discloses that it will use artificial intelligence to select candidates for interviews. Its privacy notice explains how personal information will be handled. The vendor promises faster, more consistent screening.

Then a candidate cannot complete the assessment using a screen reader. Another encounters video questions without accurate captions. A third scores poorly because the system interprets disability-related speech patterns as weak communication skills.

The employer has disclosed its technology. Has it examined whom that technology excludes?

Since Jan. 1, 2026, Ontario’s job-posting requirements generally require employers with at least 25 employees to disclose in publicly advertised job postings when they use AI to screen, assess, or select applicants. Disclosure does not establish accessibility.

Transparency is not accessibility

Employers must distinguish three questions.

• First, how does the technology collect, use, disclose, retain, and safeguard applicants’ personal information in accordance with applicable privacy rules?
• Second, do applicants know AI is being used? Ontario’s disclosure requirement addresses that question.
• Third, can candidates with disabilities use the technology and receive a fair assessment of their ability to do the job?

A system can protect information and announce its use while still excluding qualified people.

Barriers can arise at different stages. An applicant-tracking platform’s fields or navigation controls may not work with assistive technology. Such platforms do not always use AI, but an inaccessible entry point can block access to an AI-assisted assessment.

A video assessment may offer no effective captions or accessible alternative for deaf or hard-of-hearing applicants. A speech-analysis tool may mistake a stutter or atypical speech for uncertainty. A scoring model may treat disability-related differences in expression, eye contact or response time as evidence of poor performance.

Risks of hiring technologies

These are recognized risks. The U.S. Department of Justice has warned that hiring technologies, including facial and voice analysis, can screen out qualified applicants with disabilities. Its guidance concerns American law, but the practical risks deserve attention here.

Technical access and fair assessment require separate scrutiny. Someone may navigate every screen and still be disadvantaged by what the system measures.

Ontario employers already have obligations relevant to these problems. Sections 22-23 of the Integrated Accessibility Standards Regulation, made under the Accessibility for Ontarians with Disabilities Act, require notification about accommodation during recruitment and, when selected applicants request accommodation, consultation and suitable accommodation for the assessment or selection process.

The Ontario Human Rights Code also protects against disability discrimination in employment and requires accommodation to the point of undue hardship. The Ontario Human Rights Commission emphasizes inclusive design and warns that creating or perpetuating barriers may violate the Code.

Using a third-party vendor does not relieve the employer of its obligations. A vendor’s assurance of accessibility cannot establish whether the system accommodates an applicant or fairly assesses relevant abilities.

That is why accessibility must become a procurement question.

Voluntary national standard

Employers now have a Canadian benchmark to guide that work. Published in December 2025, CAN-ASC-6.2:2025, Accessible and Equitable Artificial Intelligence Systems, is a voluntary national standard covering the procurement, deployment and ongoing monitoring of AI. It calls for involving people with disabilities in procurement decisions, verifying accessibility and equity before acquisition, and monitoring actual impacts after deployment. Employers can use it as a practical due-diligence benchmark when evaluating vendors and planning oversight. Conformance, however, does not establish compliance with binding accessibility and human rights obligations; those require their own legal assessment.

Before purchasing hiring technology, employers should ask who tested it with people with disabilities, which disabilities were represented, and whether testing covered the entire process.

Employers should ask which accessibility standard, version, and conformance level were used, and what evidence supports the vendor’s claims. A technical conformance report may identify interface problems without establishing that an algorithm evaluates candidates fairly.

What happens when the system encounters disability-related characteristics it was not designed to understand? Can a reviewer identify an unreliable result? Can the assessment be changed or replaced? Can the employer meaningfully question the score?

Accommodation must be available through an accessible route outside the automated process. Candidates should not have to complete an inaccessible assessment to request an alternative. A telephone number alone will not provide effective access for everyone.

Contracts should address testing evidence, known limitations, remediation, updates, and responsibility for barriers. Employers should designate who can pause assessments or reconsider results, and how applicants can reach them.

These questions should be asked before deployment, while employers can choose what to buy and how to use it. Waiting for a complaint burdens the applicant who has already encountered the barrier.

For lawyers advising on hiring AI, the review should extend beyond data handling and disclosure to the candidate’s experience and the assessment’s validity.

An employer may be able to say, truthfully, “We told applicants we use AI.”

The consequential question is whether qualified applicants with disabilities had an equal opportunity to demonstrate what they could do.

Lorin MacDonald is a human rights and accessibility lawyer, educator, and governance advisor. She serves on Accessibility Standards Canada’s Technical Committee on Accessible Justice and is a Member of the Order of Canada.