Ontario became the first Canadian jurisdiction to mandate disclosure of artificial intelligence in hiring when amendments to the Employment Standards Act took effect on January 1, 2026. Employers with 25 or more employees must now inform job applicants when AI is used to screen their applications. The requirement marks a shift in regulatory attention toward algorithmic hiring tools, though the disclosure obligation itself does not impose testing, auditing, or accommodation requirements. Those obligations continue to flow from human rights legislation, and employers who deploy AI screening tools without understanding that framework face growing legal exposure.
The new disclosure rule is narrow in scope. It requires employers to tell applicants that AI was part of the screening process, but it does not require employers to explain how the AI system works, whether it has been tested for bias, or what recourse exists when a candidate believes the system produced an unfair result. The amendment does not require equity audits before deployment or impose any mechanism for rejected applicants to challenge algorithmic outcomes. For employers, this means the disclosure requirement is easy to satisfy but does little to address the underlying liability questions that arise when automated systems intersect with human rights obligations.
Human rights legislation in Canada has always applied to hiring decisions regardless of the method used to make them. The Ontario Human Rights Code prohibits discrimination in employment based on protected grounds including disability, race, age, and sex. The Supreme Court of Canada established decades ago that a facially neutral policy with discriminatory effects engages human rights protections. This principle applies with full force to algorithmic hiring. A machine learning system that screens out candidates with non-linear employment histories, gaps in work experience, or communication patterns that differ from the training data may produce outcomes that disproportionately affect people with disabilities, older workers, or members of racialized groups. The employer's intent is irrelevant. The question is whether the system produces adverse effects on protected grounds.
For employers considering or already using AI screening tools, the duty to accommodate adds another layer of complexity. When a job applicant discloses a disability and requests accommodation, the employer's duty to accommodate is triggered at that moment. If the accommodation request goes into an automated system that has no mechanism to process it, the employer may be found to have failed the duty to accommodate at the point of first contact. Purchasing a screening tool from a vendor does not transfer the employer's human rights obligations. The employer remains responsible for ensuring that its hiring process can accommodate applicants who disclose disabilities, and a system that cannot pause, flag, or route accommodation requests to a human reviewer creates immediate legal risk.
The vendor liability question is also evolving. In the United States, a federal lawsuit against Workday Inc. has advanced a theory that AI vendors can be held liable as third-party agents when their screening tools discriminate against applicants. The case, which alleges discrimination based on race, age, and disability, survived dismissal and was certified as a collective action in 2025. In March 2026, the court rejected the vendor's argument that age discrimination law does not cover applicants. Canadian courts have not yet addressed vendor liability for discriminatory AI tools, but the logic of the US litigation is instructive. Employers who rely on vendor assurances about bias testing may find those assurances insufficient if challenged.
In December 2025, Accessibility Standards Canada published CAN-ASC-6.2:2025, a national standard under the Accessible Canada Act addressing accessible and equitable artificial intelligence systems. The standard calls for employers to validate AI hiring tools for equitable performance across disability groups, treat statistical discrimination as a procurement risk, and provide human alternatives to automated decisions. While the standard is not itself enforceable legislation, it represents an emerging benchmark for due diligence. Employers who can demonstrate compliance with the standard are better positioned to defend against allegations that they failed to take reasonable steps to prevent discrimination. Employers who cannot demonstrate such compliance may find the standard cited against them.
The practical implications for employers using AI screening tools are significant. Before deploying any automated screening system, employers should ask whether the tool has been audited for disparate impact discrimination based on disability, race, age, and other protected grounds. They should ask whether the system can identify and route accommodation requests to a human reviewer. They should ask what recourse exists when the system produces an outcome that an applicant challenges. They should ask whether the vendor will stand behind the tool if a human rights complaint is filed. If the vendor cannot answer these questions, or if the answers reveal gaps, the employer is accepting legal risk that it may not fully understand.
Alberta employers are not yet subject to a disclosure requirement comparable to Ontario's, but the human rights framework that applies to algorithmic hiring is consistent across provinces. The Alberta Human Rights Act prohibits discrimination in employment on protected grounds, and the duty to accommodate applies to hiring processes. An employer in Alberta using AI screening tools faces the same underlying liability questions as an employer in Ontario, even without a disclosure mandate. The absence of a disclosure requirement does not create a safe harbour. It simply means the applicant may not know that AI was involved, which does not change the employer's obligations.
For employers who have already deployed AI screening tools, a compliance review is prudent. That review should assess whether the tool has been tested for bias, whether accommodation requests can be processed, and whether there is meaningful human review of automated decisions. For employers considering new tools, procurement due diligence should include these questions from the outset. The regulatory environment is moving toward greater scrutiny of algorithmic decision-making in employment, and employers who treat compliance as an afterthought may find themselves defending decisions they did not fully understand when they made them.