Evan Solomon, the federal Minister of AI, speaks at an announcement in Vars, Ont., on May 4.Spencer Colby/The Canadian Press
Labour leaders are calling on the federal government to enact legislation that would mandate employers to consult with employees and unions before they introduce artificial intelligence systems in the workplace.
A number of representatives from major unions – including Unifor, the United Food and Commercial Workers and the Canadian Union of Public Employees – met last week with Evan Solomon, the federal Minister of AI, to express concerns over the lack of worker input in Ottawa’s national AI strategy. A day later, Mr. Solomon announced that the government would create an AI and Labour Advisory Council to give workers and unions a say in shaping AI policy.
The federal AI strategy is a policy document that is currently in the works and expected to lay out a detailed plan of how Ottawa plans to govern and regulate AI technology. There is no timeline yet for the release of the document, according to public statements from Mr. Solomon. The Ministry of AI did not respond to The Globe and Mail’s questions about the strategy and the minister’s discussions with labour leaders.
Among the recommendations unions are making is to introduce new clauses in the Canada Labour Code that would compel employers to consult and be transparent with employees about what kinds of AI tools might be introduced in their workplaces so that they have a clearer understanding of if and how the technology will impact their jobs.
“The minister was very open about hearing from unions about what their members are concerned about, and my understanding is that the feds will focus on building a pro-worker AI strategy,” said Senator Hassan Yussuff, former president of the Canadian Labour Congress (CLC), who has served as an independent senator since 2021 and was present at the meeting.
Various groups including unions, academics and AI companies have warned that the technology – which is being adopted rapidly across white-collar work forces – could cause an upheaval in the labour market by transforming jobs or eliminating them entirely.
How a Toronto AI startup hopes to make the keyboard obsolete
A recent study by economists at the Brookings Institution found that AI will have an outsized impact on white-collar workers who don’t have four-year university or college degrees (usually in clerical and customer service jobs) because their career pathways to higher-wage work also risk being disrupted by AI. There could be a scenario, the authors predicted, in which tens of millions of non-university-degree workers have no stable employment and no upward mobility.
Globally, policy makers have introduced little legislation to regulate AI, including measures that would offer enhanced supports for workers in the event of mass layoffs.
“We’ve been left out of the conversation for a long time,” Bea Bruske, president of the CLC, said in a recent interview with The Globe. CLC representatives were present at the meeting with Mr. Solomon, Ms. Bruske said, and came away “encouraged” by Ottawa’s commitment to prioritize workers when thinking about how to regulate AI.
“My concern is obviously that employers will push back hard on our recommendations. Anything that limits their ability to implement something new without hurdles is not going to be received well,” she predicted.
Ottawa released its six “pillars” of AI strategy in last month’s Spring Economic Update, which included “protecting Canadians” and “powering AI adoption for shared prosperity,” but details of how these goals will be achieved have yet to be laid out.
There is some precedent, however, for the enactment of AI-specific labour legislation. The European Union’s AI Act, which took effect in 2024, decrees that employers have to inform “workers and their representatives” when deploying “high-risk” AI systems, which the act defines as systems that can pose serious risks to the fundamental rights of people. Examples would include an AI tool that retroactively identifies a shoplifter or denies an asylum claim on a discriminatory basis.
Is AI coming for our thinking? Behold the age of ‘cognitive surrender’
In Germany, employers are mandated to inform worker representatives about any planned AI deployment.
Over the past two years, Canadian unions have tried to codify language around AI and layoffs in collective agreements, going so far as to have agreements explicitly state that workers cannot be laid off because of AI technology. But they have faced pushback from employers who say that they need flexibility to implement AI technology to remain competitive.
In a submission to the federal Senate committee on social affairs, science and technology (which is assessing the impact of AI on various layers of society), CUPE has also called for the government to mandate negotiations between employers and workers when AI systems are introduced in the workplace, and enact restrictions on electronic monitoring and algorithmic decision-making in hiring, promotions, discipline and terminations.
“There have been internal discussions at CUPE about what happens if AI makes a decision on firing or disciplining a worker and how that would play out in front of an arbitrator. How would the arbitrator be able to accurately assess how the decision was made and whether there was bias or discrimination?” said Sarah Ryan, a researcher at the union who was present at the meeting with the federal Minister of AI. “What we are pushing for is laws that prohibit employers from making decisions about workers that rely on output from AI systems.”
Labour leaders also proposed that the government immediately start collecting detailed data about how AI is impacting the work force, whether in terms of changes to job functions or layoffs. The federal government’s 2025 budget allocated $25-million over six years to Statistics Canada to implement a new program – the Artificial Intelligence and Technology Measurement Program – that would track how AI is used by various organizations to understand its impact on the labour force and the economy.
StatCan currently collects some data on the extent to which AI has reduced tasks previously performed by employees, but that data set does not track job losses.