Responsible AI adoption in Canada’s public sector

AI adoption in the Canadian public sector remains uneven and fragmented. Only 22 per cent of organizations report having implemented AI tools, while another third per cent are piloting or experimenting. At the same time, nearly half of public servants report using AI in their day‑to‑day work—most often to summarize information, draft or edit documents, conduct preliminary research, and generate ideas.

This disconnect reflects a broader challenge: AI is being pulled into the workplace by employees seeking productivity gains, rather than pushed strategically by institutions with clear policies, controls, and investment plans.

Without coordinated action, governments risk creating a shadow AI environment—one that exposes sensitive data, undermines consistency, and erodes trust.

Public servants clearly see the potential of AI, but conviction remains shallow. While nearly 80 per cent say AI is important for improving productivity and operational efficiency, only a small fraction view it as “extremely important.” This suggests that many employees recognize AI’s theoretical value but have yet to see it meaningfully embedded in how work gets done.

Operational efficiency has emerged as the dominant measure of return on investment, with governments focusing AI investment on areas such as cybersecurity, finance, and administrative automation. Most expect returns within one to three years—an ambitious timeline that will be difficult to achieve without addressing foundational gaps in data quality, governance, and skills.

AI’s value proposition in the public sector is not about replacing judgment or automating policy decisions. It is about freeing capacity, improving service consistency, and enabling employees to focus on higher‑value work that directly benefits Canadians.

Few issues resonate more strongly in Canada’s public sector than data sovereignty. An overwhelming majority of public servants believe citizen data must be stored within Canada, and many fear that public trust will erode if sensitive information is subject to foreign jurisdictions.

As governments increasingly rely on cloud infrastructure and AI models, questions of control, jurisdiction, and accountability move to the forefront. When data is processed or stored outside Canada, it may be subject to foreign laws, disclosure requirements, or security risks beyond the government’s direct control.

Digital sovereignty is not about rejecting global technology providers—it is about making deliberate choices that align with Canadian values, legal frameworks, and public expectations. For AI, this means: Clear data residency requirements, strong contractual and governance controls with vendors, investment in Canadian AI infrastructure, and transparent communication with citizens about how data is used and protected.

Trust, once lost, is difficult to regain. Data sovereignty is therefore not just a technical consideration, but a cornerstone of public legitimacy.

These examples demonstrate that AI can enhance—not replace—public service, provided it is implemented responsibly and aligned with clear policy objectives.

Governments have a unique opportunity to demonstrate AI’s positive potential by leading with responsibility, transparency, and purpose.