In January 2026, a Signal49 report projected that artificial intelligence (AI) could leave Canadian employment 555,000 jobs below trend by 2030, before productivity gains push it 535,000 jobs above trend by 2045. The federal government neither contested nor endorsed the projection but promised to “monitor the impacts of potential displacements.” That single phrase captures the problem with how Canada’s AI for All strategy treats labour: confronted with the possibility of mass job displacement, the government’s committed response is to watch.
On the upside, the strategy makes concrete, funded commitments, including a $500 million Canadian Tech Growth Fund; a newly announced Sovereign Wealth Fund to back “national champions”; and goals of 250,000 added jobs by 2031 as well as an increase in AI adoption by small and medium-sized enterprises from 12 to 60 percent by 2034. These are ambitious and laudable goals. They are also dated to the early 2030s, the years Signal49 says will hurt the most, while the strategy reasons as though the 2045 recovery is assured.
What the strategy offers potentially displaced workers is categorically thinner. It promises literacy training and stakeholder alliances to “identify skills gaps.” But nowhere are there comparable funded entitlements or safety nets that pay out when workers are displaced. The strategy commits to measuring labour-market disruption through Statistics Canada, but it commits to no particular action when said measurement registers significant job loss. As it stands, the government has installed a smoke detector but has no evacuation plan.
This approach is a choice the strategy makes, and that choice rests on a bet. AI for All asserts, repeatedly, that AI is meant “to augment human expertise rather than displace it,” and the entire social architecture is built downstream of that assumption. But credible experts disagree about AI’s economic impact. Some expect rapid, possibly discontinuous change within a few years; others expect a slower trajectory where AI behaves as a conventional general-purpose technology, reshaping the tasks within jobs rather than eliminating roles wholesale.
Earlier this month, more than 200 economists and other experts — including Nobel laureates Daron Acemoglu and Simon Johnson — signed a statement warning that the economic disruption of AI could exceed the Industrial Revolution’s and urging governments to build the institutions to manage it now.
There are several live stories about AI and Canadian work, and they point in different directions. In one, AI drives large-scale displacement. In another, the real damage arrives when an overheated AI investment bubble bursts and throws people out of work for ordinary macroeconomic reasons. In a third, and one that Canada’s unions have been pressing, the threat is less about job counts than about power. In that scenario, AI is increasingly a tool that management uses to intensify surveillance, erode autonomy and weaken workers’ bargaining position.
AI for All is built for a fourth story, the relatively benign one, where adoption quietly manufactures jobs over time. But whichever of the first three unfolds, the country needs the same missing institutions: protection that activates when people are displaced, and modern rules governing how AI enters the workplace. You do not need a confident forecast to see the gap.
Although greater efforts are needed to measure the disruption effectively, the early numbers already show where the strain is landing. In one of the more rigorous US studies to date, employment for 22-to-25-year-olds in the most AI-exposed occupations has fallen by roughly 19 percent since late 2022, even as employment for older workers in the very same jobs held steady or grew. Aggregate employment has barely moved, but the damage is concentrated at the bottom of the ladder where first jobs are found. Canada shows the same shape, if not the same proof. Youth unemployment reached its highest level in over a decade in 2025 outside the COVID-19 pandemic, and the number of entry-level job postings has fallen sharply from its 2022 peak. A government waiting for the statistics to prove causation beyond dispute will be acting years after the damage has been done.
Under conditions of deep uncertainty about severe and hard-to-reverse harm, you do not wait before acting; you insure. This is the precautionary principle, and Canada already runs on it. Ottawa’s own framework directs federal departments not to treat scientific uncertainty as grounds for delay where harm could be serious or irreversible, and it is why the country manages fish stocks the way it does. That framework was not written for labour markets, but nothing in its reasoning confines it to the domains it names. The same principle should extend to the people this technology could displace.
The strategy acknowledges that we are operating under uncertainty and that revision and updating may be needed as the technology progresses. Such revision may be too little, too late, however. The mechanism that historically spread technological gains to workers was leverage — the capacity to organize, bargain and force redistribution. That leverage erodes as labour becomes easier to replace. Protections that are cheap to legislate today become more difficult once displacement is advanced and bargaining power has thinned. “Monitor and revise later” assumes a capacity to act that the disruption itself may prevent.
Two Moves Canada Can Make
Two moves, using instruments Canada already has, could help build these protections. One insures against loss; the other protects against worker disempowerment.
First, convert the proposed dashboard that monitors economic displacement into a trigger. The Statistics Canada measurement program that the strategy already funds should become the sensor for a package of protections, including wage insurance for workers who move to lower-paying roles; income support not conditional on successful retraining; and training funded as an entitlement, which activates automatically when displacement crosses defined thresholds. Some of the major labs have proposed similar ideas: OpenAI has called for expanded safety nets that activate “automatically when these metrics exceed pre-defined thresholds”; Anthropic’s Dario Amodei lists wage insurance and measurement among his recommendations. That even AI labs and their critics converge on this instrument makes Canada’s silence conspicuous.
Importantly, Canada does not have to invent the mechanism from scratch. When tariffs threatened jobs in 2025, Ottawa stood up an employment insurance pilot within weeks — waiving the waiting period, adjusting regional unemployment rates upward and adding weeks of benefits for long-tenured workers. Canada built a displacement trigger for tariffs, but AI gets a dashboard instead.
Second, act preventively and shore up worker rights while it is still possible. Start with the technological-change provisions of the Canada Labour Code, written in 1973 and never meaningfully updated for AI. Canada’s unions have asked for a real voice in workplace AI decisions and protections for workers whose conditions are reshaped by it. It is a concrete, non-radical place to begin. Only six percent of Canadian workers are federally regulated, so the code alone reaches a sliver. Protections should be extended, and the Labour Market Development Agreements are a promising lever for the rest.
If displacement runs deep, more far-reaching responses may become necessary, such as portable benefits untied from any single employer, sustained income support and the broader terrain now being mapped under the heading of “economic dignity.” Other jurisdictions are already war-gaming these scenarios; Canada should take them seriously, too.
AI for All was launched under a banner the government chose itself. Will AI improve the lives of all Canadians, or benefit only a few? The strategy has funded, precisely and in advance, its answers for the few. For everyone else, it has promised to watch — and the longer it watches, the harder it becomes to act.
