Canada tackles legacy while advancing a whole-of-government AI strategy

As the Canadian government moves to develop a whole-of-government (WOG) artificial intelligence (AI) strategy, one of the biggest hurdles it faces is its current decentralised digital governance model.

The Government of Canada’s Chief Information Officer (CIO), Dominic Rochon, notes that each department and agency has its own accountability structures.

These entities also independently seek funding, hire staff, and manage their own IT systems, which can contribute to the accumulation of technical debt.

According to a 2023 Auditor General of Canada Report, there are approximately 7,500 applications either running or in development across the federal government.

Of these, only 38 per cent have been deemed “healthy,” meaning that 62 per cent have been classified as unhealthy legacy systems requiring upgrades or replacement, notes Rochon.

This is one of the biggest challenges in developing a WOG AI strategy, as the scale and complexity of this environment make modernisation a long-term structural challenge.

Rochon tells GovInsider that AI adoption across the Government of Canada is occurring at varying levels of maturity across departments, with limited common, shared infrastructure and tools for public servants.

But the AI Strategy for the Federal Public Service is tackling some of these challenges.

He shares that the federal government’s AI Strategy is “like 1,000 flowers blooming,” in a full garden, unfurling slowly at varying speeds and stages of evolution.

The legacy of decentralisation

“The [current] decentralisation reflects long-standing administrative autonomy across ministries, which have shaped how digital systems are designed, funded, and maintained,” says Rochon.

While departments operate independently, they are required to follow a set of central policies established by the Treasury Board of Canada Secretariat, where the government CIO role sits.

These policy instruments cover areas such as government security, service and digital, privacy, and information management.

Rochon notes that this policy framework includes several overarching strategies – such as the Government of Canada’s Enterprise Cyber Security Strategy, the Government of Canada Trust and Transparency Strategy, and the AI Strategy for the Federal Public Service – supported by cascading directives and guidelines.

“Compliance is monitored centrally, but ultimate accountability remains with individual departments and agencies,” says Rochon.

Digital responsibilities within the Government of Canada are distributed across multiple organisations, each with distinct mandates.

For example, the Canadian Digital Service is supporting the development of a unified approach to online sign-in and digital credentials that will allow people to quickly and securely access federal services.

Shared Services Canada is responsible for enterprise IT infrastructure, including networks, connectivity, hardware, mobile devices, and workplace technology, and holds elements of the cyber security responsibility associated with that infrastructure.

Cyber security is further supported by the Canadian Centre for Cyber Security within the Communications Security Establishment, which provides threat monitoring, guidance, and mitigation support across government.

Responsibility for IT procurement is shared between Public Services and Procurement Canada and Shared Services Canada.

Innovation, Science and Economic Development Canada houses the Minister of Artificial Intelligence and Digital Innovation, and leads federal policy related to AI and the digital economy, while Service Canada plays a central role in delivering digital services directly to the public.

Coordinating multiple ministries

Rochon comes back to the point he made earlier: this fragmentation means that modernising IT and transforming government requires coordinating multiple ministries and departments, making governance extremely complex.

To help address these challenges, the government proposed in Budget 2025 the creation of an Office of Digital Transformation.

Rochon suggests that this office could potentially serve as a precursor to a full Ministry of Digital, similar to models adopted in other countries which have dedicated digital ministries that provide centralised technology leadership.

Whole-of-government AI tools

The Government of Canada has already introduced some WOG AI tools. Examples include GCtranslate, an AI-powered translation tool “to build a more agile, modern and bilingual public service,” says Rochon.

Ottawa is also testing another AI tool called ParlBrief that transcribes, summarises, and analyses parliamentary committee meetings. Media reports suggest that the tool could save public servants over 500 hours of work a year.

“These kinds of back-office applications could be deployed once and scaled across departments rather than duplicated multiple times,” says Rochon.

Additionally, as part of the AI Strategy for the Federal Public Service, an AI registry has been developed as a prototype. Its purpose is not to showcase achievements, but rather to provide visibility into all AI initiatives that are underway across government.

“With this registry, a centralised digital transformation office could identify duplication, spot clusters of similar needs across departments, and select a single tool to develop in partnership with a Canadian company,” says Rochon.

He notes that responsible AI adoption is an important part of the overall strategy and includes tracking where AI is used, ensuring compliance with policies, completing algorithmic impact assessments, and developing mechanisms to measure productivity gains.

“The main goals are to understand whether AI is improving efficiency for public servants, reducing wait times, improving service quality, and increasing accuracy,” Rochon highlights.

Education is a top priority, with one of the four pillars of the AI Strategy being training and talent development.

The other three are: building central AI capacity, including through establishing an AI Centre of Expertise for the Government of Canada, responsible AI adoption through clear, up-to-date legislation and policy, and lastly, greater engagement, transparency and value to Canadians.
 

The Government of Canada has partnered with Cohere to provide training and explore opportunities to deploy AI technologies to enhance public service operations, while also strengthening Canada’s commercial capacity to use and export AI.

The Canada School of Public Service has launched between 60 and 80 AI-related courses, ranging from introductory sessions to week-long immersive programmes exposing participants to AI institutes and companies.

The federal government aims to align these efforts from coast-to-coast by positioning the Canada School of Public Service as a central repository for AI training, with the potential for further support through federal–provincial–territorial collaboration.

Making AI adoption legacy-proof

“Since we know that legacy systems are often the result of flawed funding models, the question of funding becomes important here,” he says.

Departments typically seek funding from the Department of Finance during the budget process, but even when projects are approved, they often receive only partial funding.

This leaves departments frequently under-resourced from the outset, which ultimately undermines project success from the get-go, notes Rochon.

Effective modernisation requires better prioritisation, full funding, and long-term sustainability planning, he adds.

“Data is central to sovereign and responsible AI use.

“Since the Data Strategy for the Federal Public Service requires departments to curate and manage their data effectively, before any AI project is prioritised, departments must identify suitable business processes and ensure their data meets required standards,” says Rochon.

However, the challenge does not end there. “Sovereign AI within Canada’s federal government also raises broader infrastructure questions.”

Depending on hyperscalers

“Canada currently depends heavily on hyperscalers, as domestic providers have limited capacity,” says Rochon.

“Computer power and infrastructure sovereignty are being examined as part of a Pan-Canadian AI Strategy led by Minister Solomon, responsible for the AI and Digital Innovation portfolio.

This effort aligns with the broader Buy Canadian Policy, aiming to strengthen domestic capability while acknowledging current dependencies.”

Ideally, government technology should be continuously renovated in an agile manner rather than simply replaced in crisis cycles, he notes, adding that addressing legacy systems requires a stronger governance structure.

“The central challenge lies in designing a model that supports continuous modernisation, proper funding, and accountability across departments. This governance complexity lies at the heart of broader digital transformation efforts,” says Rochon.

To read our past coverage of the Canadian Digital Service, click here.