On the same day, this fall, that students at the University of Toronto were set to return to campus, over 6,100 staff were positioning themselves to strike.
The university’s Staff-Appointed Unit had spent the past three months in collective bargaining with the university’s management over their contract renegotiation, and talks had stalled over the issue of job security.
What initially caused the breakdown was U of T announcing that it would save $20-million in next year’s budget through staff reductions by not replacing employees who leave the university.
This prompted a swift backlash, and forced negotiations with United Steelworkers Local 1998, representing some of the U of T workers, to center around the issue of job security. But underpinning the stalemate was a related complication that grew more central to the negotiation over the course of the summer:
How to negotiate artificial intelligence.
In general, Canada’s unions have been pushing for more safeguards against AI, while employers have been reluctant to offer these protections saying that technology is moving too fast and the impacts are not yet known.
U of T and its staff – which include academic advisors, IT workers, program coordinators, lab assistants and even a piano technician at the music school – averted the strike at the eleventh hour and reached a tentative agreement, which has since been ratified. The new contract includes more job protections, as well as a “letter of intent” regarding AI worker protections, was among the first of its kind at U of T and in Canada, and among the first in Canada.
What this means is that although the university hasn’t offered outright protections from agents and chatbots, it did allow language into the collective agreement that would recognize a need to protect employee data, to ensure that human judgement is used to review employment decisions, to make AI training available to employees and to monitor the impact AI is having on administrative work.
“I would say that this is the first point on the board that’s going to come up again in the future,” said Feyisama Adara, who was part of the USW Local 1998 bargaining team.
“I do think that it was a huge win, absolutely, for us to be able to get some language on artificial intelligence. I do think it’s something we’re going to really understand the impact of what AI is having on our members and our workforce.”
Over the past six months, pattern bargaining among Canadian unions have introduced a growing number of clauses to address specific labour concerns around the use of algorithmic management systems, the need to protect employee data, increased workload, decreased autonomy, more surveillance and monitoring, and AI-powered human resources system.
This summer, workers at York University also demanded AI guardrails. So did workers at the Canada Revenue Agency last year. Last January, Unifor negotiated to have its members at Bell Canada set up a “Committee on Artificial Intelligence,” but also stopped short of offering actual protections.
Unions have been pushing for increased regulations for years, often negotiating general “Technological Change Clauses” into collective bargaining agreements. Technological change clauses have been more common with unions like the International Alliance of Theatrical Stage Employees (IATSE) and other unions that regularly deal with new technologies, but AI poses a new concern of unpredictability for how quickly it develops and integrates into the workplace.
The Canadian Union for Public Employees (CUPE), which also has members at U of T and helped the USW team draft the “letter of intent’ on AI has been working on AI-related issues since before 2024.
CUPE is among the most forward-thinking of the Canadian unions when it comes to addressing the rising concerns about AI, but it’s not the only one thinking ahead. The Communications Workers of America (CWA) now includes a reference to AI in the contracts it proposes to employers, like this one, which was recently proposed to the New York Times:
ARTICLE XX – ARTIFICIAL INTELLIGENCE
XX.01 The Employer and the Union recognize that artificial intelligence and machine learning technology (“AI”) could be useful tools in reporting and production but there are legitimate concerns around journalistic integrity, accuracy in reporting, integrity of the bargaining unit, and maintaining consistency with Recognition and Intent stated in Article 1.
XX.02 The Employer agrees that AI will not be used in a manner that results in the loss of employment of bargaining unit employees.
XX.03 The Employer will consult with the Union about implementing any new AI tools or processes.
Still, there remains no comprehensive, economy-wide federal law in Canada that specifically governs the use of artificial intelligence. Bill C-27, enacting the federal Artificial Intelligence and Data Act, died when parliament was prorogued in January 2025. This is another reason why union contracts, thus far, have relied on a case-by-case negotiation with employers to work out their member’s rights around AI.
“CWA Canada and many other unions always push for this kind of language protecting jobs, not
Opposing AI use or tools, but forcing discussions and actual long term planning before randomly adopting AI that may or may not work, “ wrote Carmel Smyth, President of CWA Canada, in an e-mail to rabble.
“Companies have to balance the unbridled enthusiasm for AI, with a reality check,” she wrote.
“Many employers are resistant to including any language related to adopting AI in their collective agreements. They take the position that they don’t know what the possible applications are yet.
We (CWA Canada and many unions) believe employers are thinking that one day AI will be able to eliminate some human work and they don’t want to be behind that trend,” she continued.
“Our position is that employers can introduce all the AI they want, they just can’t use it to cut jobs.”


