Canadian tech firms face AI bias gap in HR policies

QueerTech has published a report on the use of artificial intelligence in human resources across Canada's technology industry. The study found a wide gap between reported confidence in inclusive practice and the policies and safeguards in place.

The national non-profit said the research is the first in Canada to assess readiness to implement, manage and govern responsible AI across HR functions in the tech sector, with a focus on outcomes for 2SLGBTQI+ professionals.

The findings suggest AI is already deeply embedded in people management. According to the report, 91% of Canadian tech companies use AI in at least one HR function, with or without human oversight, while 61% fully automate at least some HR decisions without human oversight.

Use is especially notable in hiring and employee monitoring. More than half of companies use AI in specific recruitment functions, and nearly 60% use it in some employee monitoring and management tasks.

The report also points to further expansion in more sensitive areas. HR professionals surveyed said 43% of Canadian tech companies are expected to use AI for automated termination and offboarding functions by 2029.

Bias concerns

Despite that level of adoption, the research found strong concern among HR staff about the risks attached to these systems. Some 89% of HR professionals said they were concerned about potential bias in AI-guided training, assessment and wellbeing tools, while nearly half said accuracy and reliability were their main concern in AI-powered HR systems.

The report draws a distinction between concern about AI and confidence in inclusion practices. It found many HR professionals rated their own knowledge highly even where workplace measures appeared limited.

For example, 62% of HR professionals said they were very confident in their knowledge of inclusive hiring processes, yet only 51% of Canadian tech companies reported having comprehensive policies that address queer representation.

On workplace support, 63% said they were very confident in their knowledge of designing policies to support queer employees. Yet fewer than 46% of companies offer same-sex partner benefits, gender-affirming care coverage, chosen family care leave, or surrogacy and adoption support.

The same pattern appeared in questions about workplace culture. While 58% of HR professionals said they were very confident in their ability to build allyship and awareness across teams and managers, 80% believed 2SLGBTQI+ people were well represented at their workplace and 86% believed queer employees were out or comfortable being out.

Another result pointed to gaps in practical understanding. Although 59% said they were very confident in their knowledge of 2SLGBTQI+ concepts and terminology, 87% also said they needed 2SLGBTQI+-specific tools, resources and training materials.

Hostile responses

The research also recorded discriminatory responses from participants. QueerTech said 10% of responses submitted by Canadian technology companies were homophobic, transphobic or otherwise queerphobic in tone.

That finding adds to concerns about how workplace systems are designed, governed and used, particularly when organisations rely on data collection, internal assessments and automated decision-making in high-risk employment areas.

The report follows an earlier QueerTech study on inclusive AI development and deployment. In that research, 97% of tech companies said inclusive systems development was a high priority, but fewer than half believed their AI products met the needs of 2SLGBTQI+ users.

Taken together, the two studies argue that risks linked to exclusion are not resolved once AI tools move from product teams into workplace use. Instead, the report suggests those risks can spread across recruitment, benefits, policy creation, monitoring and dismissal, where governance may be weaker or more fragmented.

The project received support from Women and Gender Equality Canada.

A brief context statement accompanied the findings from QueerTech's leadership.

"Our goal with this research is to give Canada's tech sector a clear baseline for where we stand, and the results show we have significant work to do. True innovation requires responsible systems implementation, monitoring and governance. If we do not actively audit and manage the AI systems dictating who gets hired, fired or forgotten, we risk automating discrimination at scale," said Naoufel Testaouni, Co-Founder and Chief Executive Officer of QueerTech.

"We are calling on people teams and tech executives nationwide to use this data to bridge the gap between intent and impact. These findings are further evidence that Canada, and our national innovation economy, cannot afford the cost of exclusion or irresponsible technologies," Testaouni said.