AI in hiring: legal risks under Ontario's Human Rights Code

Artificial intelligence (AI) has quickly become a fixture in the hiring process across many industries. Employers are increasingly adopting algorithmic tools and automated decision-making systems to screen resumés, assess video interviews, and even predict candidate performance or culture fit. While these innovations offer potential efficiency gains and data-driven insights, they also introduce serious legal risks, particularly in the context of Ontario’s Human Rights Code. 

The code prohibits discrimination in employment based on specific protected grounds. When AI systems are deployed in hiring, they can inadvertently perpetuate or amplify biases embedded in their training data or algorithms, raising the risk of systemic discrimination. Employers using or considering AI must ensure that these tools comply with legal standards, or they may face human rights complaints and reputational damage. 

This article examines the use of AI in hiring, the potential risks under the code, and best practices for employers in Ontario to navigate this evolving legal landscape. 

The rise of AI in recruitment 

AI tools in hiring take various forms. Some systems utilize natural language processing to evaluate resumés or cover letters, identifying keywords that align with job requirements. Others analyze video interviews, using facial recognition or voice pattern analysis to rate a candidate’s emotional expression, confidence, or honesty. Some platforms claim to predict workplace performance or the likelihood of success based on complex data sets. 

Large employers and human resources platforms are already integrating these tools into early-stage recruitment workflows, particularly to handle high volumes of applicants. For example, AI might automatically reject resumes that don’t meet specific criteria or rank candidates based on historical data. 

While these systems promise efficiency, their algorithms are often opaque and based on historical “successful” hire patterns. 

Legal framework around human rights 

Under the code, employers have a legal obligation to provide equal employment treatment without discrimination based on protected grounds. These grounds include race, sex, age, disability, sexual orientation, gender identity, marital status, family status, ethnic origin, place of origin, religion, and other protected characteristics. 

This obligation applies at all stages of employment, including recruitment, screening, interviewing, and the selection process. The employer may be liable for discrimination if a candidate is mistreated because of a protected characteristic. 

Importantly, intent is not required to establish a breach of the code. Discriminatory outcomes resulting from neutral policies or practices, including AI, can constitute “adverse effect discrimination” if they disproportionately impact a protected group. 

How AI can perpetuate discrimination 

AI is not inherently discriminatory. However, it is only as objective as the data on which it is trained and the assumptions underlying its design. Many AI systems rely on machine learning, where algorithms are trained using historical hiring data. If past hiring decisions reflected biases, such as favouring male candidates over female ones or prioritizing applicants with Anglo-sounding names, the AI system can learn and replicate those biases. 

For instance, Amazon famously abandoned an AI recruitment tool after discovering that it systematically downgraded resumés containing the word “women” because it had been trained on resumés submitted to the company over a 10-year period, predominantly from men. 

AI systems that assess facial expressions or vocal tone may also disadvantage people with disabilities, neurodivergent traits, or language accents. Video analysis algorithms have also been shown to struggle with accurately identifying facial expressions in individuals with darker skin tones, which may lead to skewed assessments of Black or Indigenous candidates. 

Even if the algorithm does not explicitly consider protected characteristics, its design or data inputs may produce outcomes that correlate with those characteristics. For example, a resumé screening tool that favours postal codes associated with high-income neighbourhoods may indirectly disadvantage racialized or immigrant applicants who live in lower-income areas. 

Employer liability under the code 

In Ontario, employers can be held legally responsible for discriminatory hiring practices resulting from the use of AI, even if a third-party vendor developed the algorithm. Under the code, an employer cannot delegate or outsource human rights obligations. If an AI tool results in discrimination, the employer, not just the tech provider, can be named in a human rights complaint. 

The Ontario Human Rights Tribunal (HRTO) considers whether the employer took reasonable steps to prevent discriminatory outcomes. Ignorance of how an algorithm works, or blind reliance on a vendor’s claims of fairness, is unlikely to provide a successful defence. 

While the HRTO has not yet issued a significant decision directly addressing AI-driven hiring discrimination, it has been clear in previous rulings that systemic practices leading to disparate outcomes can violate the code. As AI systems become more prevalent, it is only a matter of time before these issues are tested in formal complaints. 

Data privacy and transparency concerns 

The risks posed by AI in hiring extend beyond discrimination. There are also significant concerns related to transparency, informed consent, and data privacy. These issues intersect with Ontario’s privacy laws and emerging federal legislation, such as aspects of the previously proposed Artificial Intelligence and Data Act (AIDA). 

Job applicants are rarely informed about how their personal data is being analyzed or the algorithm's evaluation criteria. Unlike human interviewers, AI systems do not provide explanations for their decisions, making it difficult for candidates to challenge or appeal unfair assessments. 

This lack of transparency makes it more difficult for individuals to determine whether they’ve been discriminated against, which may prevent them from asserting their rights under the code. It also undermines accountability since the decision-making process may be proprietary or protected by trade secrecy claims. 

Employers can mitigate the legal risk of AI recruitment 

Ontario employers who use or consider AI-based hiring tools must take proactive steps to ensure they are not exposing themselves to legal liability. The following measures are essential. 

First, employers should thoroughly vet any AI tools before deployment. This includes demanding transparency from vendors about how the algorithm was developed, what data was used to train it, and the steps taken to mitigate bias. Independent audits, bias testing, and certifications may be helpful, but due diligence should not stop there. 

Second, human oversight must remain a central part of the hiring process. AI should not be used to decide who is hired or rejected. Instead, it should support, not replace, human judgment. Employers must ensure that trained hiring personnel review AI-generated rankings or assessments and have the authority to override them as needed. 

Third, employers should document their policies and practices regarding the use of AI, including how they assess and mitigate bias. This includes maintaining records of how AI tools are selected, tested, and ultimately decided. A paper trail can demonstrate that the employer took reasonable steps to comply with its obligations under the code. 

Fourth, employers should inform applicants when AI is used in the hiring process. While not currently required by law, this transparency supports fairness and can help prevent legal disputes. Informed applicants are better positioned to raise concerns, request accommodation, or seek clarification if they feel disadvantaged by the process. 

Employers should consult legal counsel when integrating AI into their hiring practices. Ontario’s human rights laws are complex, and the legal landscape surrounding algorithmic decision-making is still evolving. Legal counsel can help employers develop compliant policies, review contracts with vendors, and respond effectively if a complaint is filed. 

Regulation of AI use and employer compliance 

As AI continues to reshape the workforce, governments worldwide are beginning to regulate its use in employment. Ontario employers should look for guidance from the Ontario Human Rights Commission, which has previously emphasized the importance of preventing systemic discrimination in technology. Future policy statements or tribunal decisions will likely provide more direction on how AI can be used responsibly in hiring. 

Until then, the safest approach is to treat AI as a tool, not a decision-maker, and apply the same human rights principles that govern traditional hiring processes. Equity, transparency, and accountability must remain at the core of recruitment practices, regardless of whether human or machine-made decisions are involved. 

Using artificial intelligence in hiring presents both opportunities and serious legal challenges. In Ontario, employers must ensure that any AI systems used in recruitment comply with the code, particularly concerning discrimination based on protected grounds. Failure to do so can result in legal liability, reputational harm, and a loss of trust among candidates and employees. 

Employers should approach AI adoption cautiously, seek legal advice, and implement safeguards to ensure fairness and compliance. Ultimately, responsible use of AI in hiring requires technological sophistication and a deep commitment to human rights and equality in the workplace. 

Paulette Haynes is the founder of Haynes Law Firm, a boutique employment law firm in Toronto.