AI in Hiring Canada: The Rules BC Employers Miss

Resume screening was one of the first places small businesses let software make a call about a human being. That is fine right up until someone asks how the call was made. If you run a shop in Burnaby or a services firm in Vancouver and you have an AI tool ranking applicants, the honest questions are: what rules apply to you today, and what would you say if a rejected candidate asked for an explanation?

Here is what the primary sources actually say about AI in hiring Canada wide, and where they are silent.

AI adoption is real, but nobody is measuring hiring specifically

Statistics Canada found that 19.2% of Canadian businesses used AI to produce goods or deliver services in the 12 months before the second quarter of 2026, up from 6.1% in the second quarter of 2024 (Statistics Canada). Adoption skews urban at 21.0% versus 9.9% rural, and the top reported uses were data analytics at 36.6%, text analytics at 34.5%, and virtual agents or chatbots at 28.2%.

Worth naming the gap: that survey does not break out recruiting or HR as a use case. So anyone quoting you a hard number for "Canadian businesses screening resumes with AI" is guessing. What we can say is that text analytics is the second most common application, and resume screening is text analytics.

Ontario now requires disclosure. BC does not, yet.

Ontario is the bellwether. Since January 1, 2026, a publicly advertised job posting under Ontario's Employment Standards Act must include a statement disclosing the employer's use, if any, of AI to screen, assess or select applicants (Government of Ontario). The definition of AI in that guidance is broad: "a machine-based system that, for explicit or implicit objectives, infers from the input it receives in order to generate outputs such as predictions, content, recommendations or decisions that can influence physical or virtual environments."

Read that again if you assumed keyword filters are exempt. A ranked shortlist is a recommendation.

The same Ontario rules require expected compensation or a range, ban Canadian experience requirements, require you to say whether a vacancy actually exists, require notifying interviewed applicants within 45 days of the final interview, and require keeping postings and application forms for three years. The threshold is 25 employees on the day the posting goes up.

British Columbia has no equivalent AI disclosure rule as of September 2026. That matters less than it sounds, because plenty of Metro Vancouver employers post nationally, and an Ontario-facing posting pulls you into Ontario's rules.

What already binds you in British Columbia

The rules that govern AI in hiring in Canada mostly predate AI. Three BC instruments do the real work here, and none of them mention it by name.

The Human Rights Code prohibits publishing an employment advertisement that expresses a limitation, specification or preference based on a protected characteristic, unless it is a bona fide occupational requirement. The BC Human Rights Tribunal sets out the BFOR test as three parts: a legitimate job-related purpose, good faith adoption, and reasonable necessity such that the employer could not accommodate without undue hardship (BC Human Rights Tribunal). A model that systematically downranks a protected group produces the same outcome as a discriminatory ad, with worse documentation.

The Pay Transparency Act, section 2, requires employers to specify the expected salary or wage, or a range, in an advertisement for a publicly advertised job opportunity. Section 3 prohibits seeking an applicant's pay history unless it is publicly accessible (Pay Transparency Act). If your sourcing tool scrapes salary history, that is a problem in BC.

The Personal Information Protection Act is the one people forget. PIPA section 35 requires that when personal information is used to make a decision that directly affects an individual, the organization retain it for at least one year after using it, so the person has a reasonable opportunity to get access. Section 23 gives individuals the right to their personal information, how it has been used, and who it was disclosed to (PIPA). Applicants are not employees, so the section 13 employee exception does not carry you.

Practically: if your AI scored a candidate, that score is personal information used to make a decision about them, and they can ask for it.

The bias evidence is not hypothetical

In a peer-reviewed audit, Wilson and Caliskan tested text embedding models on over 500 real resumes and 500 job descriptions across nine occupations. The models significantly favoured White-associated names in 85.1% of cases and female-associated names in only 11.1% of cases, and Black male-associated names were disadvantaged in up to 100% of cases (Wilson and Caliskan, 2024).

That study tested retrieval and embedding models, not any specific commercial applicant tracking system, so don't read it as a verdict on your vendor. Read it as the base rate you are working against when nobody audits.

What Canada's privacy regulators expect

The federal, provincial and territorial privacy authorities published joint principles for generative AI in December 2023 (Office of the Privacy Commissioner of Canada). Nine principles, and three of them land squarely on hiring. Openness asks you to clearly communicate whether an AI tool will be used in a decision-making process and in what capacity. Accountability asks you to run assessments such as Privacy Impact Assessments. And the document asks that affected individuals get an effective challenge mechanism, with the opportunity to request human review or reconsideration.

These are guidance, not statute. They are also exactly what a regulator will hold up when something goes wrong.

For a picture of what a mature standard looks like, the federal government's own Algorithmic Impact Assessment is a 106 question tool, 65 risk questions and 41 mitigation questions, that sorts automated decision systems into four impact levels from Level I to Level IV (Treasury Board of Canada Secretariat). It binds federal institutions, not your bakery. But the shape is borrowable, and no private-sector equivalent is required in BC.

A short checklist that covers most of it

• Say it in the posting. One clear line that AI is used to screen or assess applications. Required in Ontario, cheap insurance everywhere else.
• Never let the model send the rejection. Keep a person on every no, and record who.
• Keep the scores. PIPA's one year retention on decision information is a floor, not a target, and Ontario wants postings and applications kept three years.
• Test for adverse impact before launch and quarterly after. Compare pass rates across groups on your own historical applicant pool.
• Publish the pay range and stop asking for pay history. Both are BC law already.
• Write down what the tool does, in plain language, before a candidate or a regulator asks.

Where this doesn't apply

If you get 12 applicants for one role, AI screening is not worth the compliance surface. Read them yourself. The economics of automated screening only work at volume, and most Burnaby and New Westminster small businesses are not at that volume for hiring even when they are for customer support.

If you can't audit it, don't ship it. A vendor who won't give you scoring outputs or per-group pass rates has handed you a liability you cannot inspect. That is a reason to keep the tool in a sourcing or scheduling role, not a screening one.

And the sources genuinely diverge on scope. Ontario legislated disclosure. BC has not. The privacy commissioners ask for human review and challenge mechanisms, but that is guidance. The federal Algorithmic Impact Assessment is mandatory only inside government. Anyone who tells you there is one clean Canadian rulebook for AI in hiring is selling something.

The safest place for AI in a small hiring process is upstream of the decision: drafting the posting, answering candidate questions, scheduling interviews, chasing references. Those save real hours and put nobody in front of a tribunal.

Sources

• Government of Ontario, "Requirements related to publicly advertised job postings," Your Guide to the Employment Standards Act, 2026. https://www.ontario.ca/document/your-guide-employment-standards-act-0/requirements-related-publicly-advertised-job
• Statistics Canada, "Analysis on artificial intelligence use by businesses in Canada, second quarter of 2026," 2026. https://www150.statcan.gc.ca/n1/pub/11-621-m/11-621-m2026010-eng.htm
• Office of the Privacy Commissioner of Canada and provincial and territorial counterparts, "Principles for responsible, trustworthy and privacy-protective generative AI technologies," 2023. https://www.priv.gc.ca/en/privacy-topics/technology/artificial-intelligence/gd_principles_ai/
• Province of British Columbia, Pay Transparency Act, SBC 2023. https://www.bclaws.gov.bc.ca/civix/document/id/complete/statreg/23018
• Province of British Columbia, Personal Information Protection Act, SBC 2003 c. 63. https://www.bclaws.gov.bc.ca/civix/document/id/complete/statreg/03063_01
• BC Human Rights Tribunal, "Employment advertisements and discrimination." https://www.bchrt.bc.ca/human-rights-duties/employment/advertisements/
• Wilson, K. and Caliskan, A., "Gender, Race, and Intersectional Bias in Resume Screening via Language Model Retrieval," arXiv:2407.20371, 2024. https://arxiv.org/abs/2407.20371
• Treasury Board of Canada Secretariat, "Algorithmic Impact Assessment tool," Government of Canada. https://www.canada.ca/en/government/system/digital-government/digital-government-innovations/responsible-use-ai/algorithmic-impact-assessment.html

If you're weighing where AI belongs in your hiring process, or you just want a second read on a tool a vendor is pitching you, book a free call with Autana Solutions. We build AI employees for Vancouver area businesses, and we'll tell you plainly when the answer is to leave a step human.