How many Canadian federal public service workers can AI disrupt, and where.
AI in the federal public sector
The Canadian federal government, under its new AI drive, plans to use artificial intelligence more broadly throughout public service operations. Stephen Burt, the Chief Data Officer, has signaled that the adoption of AI may lead to some job cuts, though he does not specify how many or in which departments. The government’s goal is also to reduce program spending by about 15 percent over three years, which adds pressure to find efficiencies.
Canada’s public workforce is large, complex, and increasingly exposed to AI. According to recent research, nearly three-quarters of public sector jobs are in occupations that are “highly exposed” to AI technologies. Among those, a substantial share are in jobs where AI could substitute core tasks rather than merely assist.
Federal employees account for about 405,900 of all public sector workers (which includes provincial, territorial, and municipal levels). Across all levels, over 1.1 million people work in public-sector roles in Canada.
Here are estimates and scenarios for how many jobs might shift, be transformed, or even disappear, especially in particular departments, over the next 3-5 years.
Baseline Risk and Exposure
Research by Statistics Canada and related studies suggests that about 60 percent of Canadian employees are in jobs that may face AI-driven transformation, some tasks replaced, others augmented. In the public sector, the risk is somewhat higher: about 74 percent of public-sector roles are in “high exposure” occupations. Of those, many are in jobs where the new technologies are likely to substitute for some tasks, not just assist.
“High exposure, low complementarity” roles, meaning jobs where AI could take over many core tasks, make up a large share in federal public service, especially in administration, finance, and clerical occupations.
Size and Scale of Key Departments
To understand potential job losses or transformations, it helps to see how large some of the relevant departments are:
The Canada Revenue Agency (CRA) employs about 59,155 people.
Immigration, Refugees and Citizenship Canada (IRCC) has a total workforce of about 10,900 to 11,000 (depending on what categories and contracts are included) in “substantive” positions.
The entire federal public service (core administration plus agencies) has about 367,772 employees.
Estimated Impacts: 3-5 Year Forecast
Given those numbers, here’s what a plausible scenario looks like:
If roughly 58-60% of CRA’s 59,000 employees are in roles with high exposure, that’s about 34,000 to 36,000 workers whose jobs are at risk of significant transformation. Of those, perhaps 20,000-25,000 may see task substitution to the point where job roles are dramatically changed or eliminated. The remainder might see augmentation; human plus AI.
In IRCC, with ~11,000 workers, perhaps 6,000 to 7,000 are in high-exposure roles. Of those, maybe 3,000-4,000 may be in low-complementarity roles (i.e. where AI replaces more than assists). The rest will likely see their roles shift; more oversight, more dealing with exceptions, less routine processing.
For the federal public service as a whole (≈ 368,000 employees), if 74 percent are in high exposure, that means around 270,000 people are likely to see their work affected by AI in the next few years. Of those, perhaps half (≈ 130,000-150,000) are in roles where AI could substitute core tasks rather than just assist. Those roles would be at most risk of job loss or major transformation.
Which Job Types and Roles Within Departments Will Suffer Most
The departments where the above numbers cluster are those with lots of:
Clerical and administrative tasks: data entry, form processing, routine correspondence, and standardized decision processes (e.g. initial screening of applications).
Customer service or public-inquiry systems (call centres, inquiries by email or chat) especially for high-volume, low-complexity questions.
Initial review or screening roles (immigration, customs, tax) where many cases follow standard rules.
Finance, procurement, and contract review for routine, repeated contracts or purchasing tasks.
So within CRA, many administrative assistants, junior auditors, verification staff, compliance clerks, etc., will be more vulnerable. Within IRCC, case processors who handle standard application flows will be most exposed. In Service Canada, benefit adjudicators or first-contact benefit and eligibility workers may see large shifts.
What is Less Likely to Be Lost: Roles That Will Shift Instead
Certain roles are less likely to disappear but will change significantly. These include:
Positions requiring discretion, nuanced judgment, or empathy (social work, certain regulatory work).
Senior policy, strategic planning, legal, or advisory roles.
Front-line services needing in-person presence or contextual decision-making.
These jobs will often see AI tools integrated as assistants: to generate reports, to analyze data, to support decision-making. The human component remains crucial.
Caveats, Timing, and Uncertainty
A few important caveats:
Exposure does not equal immediate job loss; legal, institutional, financial, ethical constraints slow change. Many “at risk” jobs instead will have their tasks transformed rather than wholly eliminated.
There is often a lag: technology deployment, training, procurement, oversight, and regulation all take time.
Some departments may resist or adapt more slowly; union agreements, public expectations, and policy scrutiny will slow sharp cutbacks.
New jobs will also be created: oversight, AI ethics, model auditing, programming, data cleaning, supervision of AI, etc. However, the skills required may differ, and not everyone displaced will easily move into those.
Bottom Line
Over the next three to five years, it is plausible that tens of thousands of Canadian federal public service jobs will either be significantly transformed or lost, particularly in departments with large administrative, clerical, or standardized processing functions. In CRA alone perhaps 20,000-25,000 jobs might face deep transformation; in IRCC perhaps 3,000-4,000; across federal public services perhaps 100,000-150,000 workers in roles where AI could substitute core tasks.
For many, change won’t be sudden but will feel inevitable: shifting task mixes, greater reliance on monitoring and oversight of AI, fewer purely routine jobs. The question for Ottawa is how well it manages retraining, redeployment, ethical oversight, and transparency so that the transformation strengthens public service instead of destabilizing it.