Ontario employers who use Artificial Intelligence (AI) to screen resumes are fully liable if the software discriminates against candidates based on age, race, or gender. Under the Ontario Human Rights Code, you cannot blame the algorithm; failing to audit your AI for built-in bias can result in Human Rights Tribunal of Ontario (HRTO) complaints and severe financial penalties.
The recruitment landscape in Ontario is evolving rapidly. From massive financial institutions in Toronto to tech startups in Waterloo, employers are increasingly relying on Artificial Intelligence (AI) to screen thousands of resumes in seconds. 🤖 While these automated applicant tracking systems (ATS) save time, they carry a hidden and massive legal risk: baked-in algorithmic discrimination.
AI software learns from historical hiring data. If a company historically favored younger candidates from specific universities, the AI will quietly filter out older applicants or individuals with foreign work experience. The Ontario Human Rights Code explicitly prohibits discrimination in the hiring process based on protected grounds like age, race, ethnic origin, and gender. If your AI screening tool illegally filters out a candidate, the Human Rights Tribunal of Ontario (HRTO) will hold your company responsible, not the software developer. In this guide, we will warn Ontario employers about AI hiring risks and how to ensure legal compliance.
Step-by-Step Guide for Employers to Prevent AI Hiring Bias
Proactive compliance is cheaper than defending a human rights complaint. It is highly advised for HR departments to consult an employment law firm from our directory to review their AI usage policies. 💼
Step 1: Audit the AI Tool Before Implementation
Before deploying any AI software to sort resumes, your HR team must demand transparency from the vendor. Ask the software provider for an algorithmic bias audit. You need to know exactly what data points the AI uses to rank candidates. If the AI analyzes graduation dates to estimate an applicant’s age and filters out older workers, you are immediately in violation of Ontario law.
Step 2: Remove Protected Grounds from the Algorithm
Ensure the AI is programmed to ignore any data related to the protected grounds under the Human Rights Code. 📝 The system must be blind to names (which can indicate race or gender), graduation years (which indicate age), and postal codes (which can be a proxy for socioeconomic status or ethnic neighborhoods in cities like Brampton or Markham).
Step 3: Establish Clear, Objective Hiring Criteria
The AI should only evaluate candidates based on objective, bona fide occupational requirements. For example, the algorithm should scan exclusively for specific certifications (like a CPA license in Ontario), years of relevant experience, and hard technical skills. Document these exact criteria in your HR policy so you can defend your hiring process if a rejected candidate files a complaint.
Step 4: Provide Human Oversight for Rejected Candidates
Never let the AI have the absolute final say. Implement a human review process where an HR professional periodically audits a random sample of resumes that the AI rejected. 👤 If the human reviewer notices that the software is disproportionately rejecting candidates of a certain demographic, the AI must be paused and recalibrated immediately.
Step 5: Update Your Workplace Privacy and AI Policies
As of January 1, 2026, under the Employment Standards Act (ESA) and O. Reg. 476/24, Ontario employers with 25 or more employees must disclose the use of AI in any publicly advertised job posting if they use it to screen, assess, or select applicants. While smaller employers (with fewer than 25 employees) are currently exempt from this job-posting disclosure duty, updating your privacy policy and career portal to be transparent about AI resume screening is still a best practice to avoid potential liability under the Human Rights Code.
Step 6: Respond Promptly to Human Rights Complaints
If a rejected candidate files an HRTO application claiming algorithmic bias, take it seriously. Do not simply reply, “The computer made the decision.” 💬 Your lawyer will need to prepare a formal Response (Form 2) proving that your AI tool was unbiased, regularly audited, and strictly evaluated candidates based on objective skills.
How Much Does Algorithmic Discrimination Cost Employers?
Failing to monitor your AI software can lead to significant legal and financial consequences for an Ontario business.
- AI Audit Fees: Hiring an independent tech consultant or law firm to audit your HR software for bias generally costs between $2,000 and $5,000 CAD.
- Defending an HRTO Claim: Retaining a lawyer to defend your company against a discrimination claim at the Tribunal typically costs $10,000 to $20,000 CAD.
- Financial Damages: If the HRTO finds your AI was discriminatory, they can order you to pay the applicant for lost wages and injury to dignity, which often results in settlements ranging from $5,000 to $25,000 CAD per affected candidate.
How Long Does the Process Take?
A candidate who suspects they were illegally filtered out by your AI has one year from the date they were rejected to file a complaint with the HRTO. If a complaint is filed, the Tribunal process-including mandatory mediation and a potential public hearing-will force your company into a legal battle that can last 1 to 2 years.
AI Bias Risks vs. Compliance Strategies
| Graduation Year Filtering | Direct Age Discrimination | Program AI to ignore dates; focus only on total years of relevant experience. |
| Name & Language Analysis | Race & Ethnic Origin Bias | Implement “blind” resume parsing where names are redacted before AI ranking. |
| Postal Code Profiling | Indirect Systemic Discrimination | Remove geographic location sorting unless it strictly relates to bona fide travel requirements. |
Frequently Asked Questions (FAQ)
Can the software developer be held liable instead of my company?
No. Under the Ontario Human Rights Code, the employer is ultimately responsible for the hiring process. You cannot outsource your human rights obligations. If the tool you purchased is biased, your company will pay the damages.
Do we have to tell applicants we are using AI?
Yes, if you employ 25 or more employees on the day a job is posted. Under O. Reg. 476/24 of the Employment Standards Act, employers of this size are legally required to disclose in any publicly advertised job posting if they use AI to screen, assess, or select candidates. Smaller businesses with fewer than 25 employees are exempt from this specific requirement, but they must still comply with human rights laws.
How does an applicant even know the AI was biased?
Applicants often run tests by submitting two nearly identical resumes-one with a younger-sounding profile and one with an older profile. If the AI systematically rejects the older profile, they have strong grounds for an HRTO complaint.
What is a ‘bona fide occupational requirement’ (BFOR)?
A BFOR is a legally allowed requirement that might otherwise seem discriminatory, but is absolutely essential for the job. For example, requiring a valid driver’s licence is a BFOR for a delivery driver, even if it filters out some applicants.
Leave a Reply