Government Files is The Canada Report’s public-records analysis series examining government documents obtained through Canada’s Access to Information (ATI) and provincial Freedom of Information (FOI) laws. These transparency laws allow members of the public to request internal government records from federal and provincial institutions. This article reviews documents released through those processes and summarizes what the records contain and what they show. While we strive for accuracy, this article represents an analysis and interpretation of the source material. For complete accuracy and full context, readers should review the original documents, which are available in full below.
Full Document
The complete document is available for download below:
As federal departments grapple with how to safely integrate generative artificial intelligence into their operations, internal government records reveal that Shared Services Canada is developing its own secure AI chatbot solution for use across the public service. The tool, called CanChat, is being positioned as a made-in-Canada alternative that could provide federal employees with AI assistance while maintaining elevated security and privacy protections.
The records, obtained through an Access to Information request to Invest in Canada, consist of an internal communiqué distributed across the Government of Canada in January 2025 outlining the federal approach to generative AI adoption.
What the Documents Show
The January 22, 2025 communiqué from Shared Services Canada provides a snapshot of how the federal government is approaching the rapid adoption of generative AI tools. According to the document, SSC and the Treasury Board of Canada Secretariat are jointly leading efforts to ensure these tools are deployed safely, securely, and responsibly across the Government of Canada.
The most notable revelation is the development of CanChat, described as a GenAI solution that SSC continues to refine. The document states that this tool may support the business of the Government of Canada while providing an elevated level of safety and security compared to commercial web-based AI solutions. SSC indicated it would release more information about the solution and its general availability over the coming fiscal year.
As Ottawa moves forward with the development of its AI chatbot, CanChat, it underscores the importance of transparency in government operations, a topic we explored in our introduction to Government Files, which delves into the public records that shed light on such initiatives.
As federal departments explore the integration of generative artificial intelligence, the recent move by Loblaw to incorporate ChatGPT into its PC Express service for enhanced grocery shopping provides a relevant example of how businesses are adapting to this technology. For further context on AI applications in the retail sector, see our coverage of Loblaw's initiative here.
The communiqué also outlines a broader procurement strategy for generative AI tools. SSC planned to issue an Invitation to Qualify on CanadaBuys in early 2025 as a first step toward acquiring GenAI tools for government back-office productivity. The stated goal was to establish procurement vehicles within twelve months, enabling departments to purchase AI tools that offer different options at various price points to support Microsoft 365 productivity improvements.
Want future Government Files like this?We release new Canadian public-records breakdowns weekly.
The Current Framework for AI Use
The records reveal that the federal government has already established a policy framework for AI use. The Treasury Board has issued a Microsoft Copilot for Work Policy Implementation Notice providing guidance on safe and secure use of existing generative AI tools. For departments choosing to use Microsoft Copilot (formerly known as Bing Chat Enterprise), SSC recommends implementing EntraID within their Microsoft 365 tenant, a feature included in E3 and E5 licences that provides additional data and privacy protections.
Additionally, SSC's Artificial Intelligence Centre of Excellence, working with TBS, has developed a Guide on the Use of Generative AI containing principles and best practices for both employees and organizations deploying these tools.
Departmental Responsibilities and Data Governance
The communiqué places significant responsibility on individual departments for AI governance. Each departmental Chief Information Officer is responsible for ensuring that guidance and policies related to GenAI tools are in place within their organization, as outlined under the Policy on Service and Digital.
The Treasury Board was also preparing additional guidance to clarify roles and responsibilities for AI adoption across departments. This forthcoming guide would emphasize the importance of data governance and the involvement of Chief Data Officers, while highlighting the need for departments to assess their data maturity and capacity before implementing AI projects. The focus on alignment with privacy, security, and legal standards suggests the government is taking a cautious approach to AI integration, prioritizing risk management alongside productivity gains.
What's Not in the Documents
While the communiqué provides a high-level overview of the government's AI strategy, several important details are absent. There is no indication of the budget allocated for either the CanChat development or the broader GenAI procurement initiative. The document does not specify which departments are working with SSC to define business requirements for the procurement, nor does it provide technical specifications for CanChat or explain how it differs architecturally from commercial alternatives like Microsoft Copilot.
The records also do not reveal the current uptake of generative AI tools across government departments, though the document notes that interest in GenAI tools continues to grow across the Government of Canada. Any security assessments, privacy impact analyses, or risk evaluations that presumably informed this approach are not included in this release.
Broader Implications
The development of CanChat reflects a broader tension in government technology procurement between leveraging commercial solutions and maintaining sovereignty over sensitive data. By developing an in-house AI solution, the federal government appears to be hedging its bets, ensuring it has an option that keeps data within government-controlled infrastructure while still enabling departments to purchase commercial alternatives through the planned procurement vehicles.
The emphasis on data maturity assessment before AI implementation suggests the government recognizes that successful AI deployment requires more than simply purchasing tools. Departments will need to evaluate whether their data governance practices, workforce skills, and IT infrastructure are prepared for AI integration, potentially slowing adoption but reducing the risk of problematic deployments.
The twelve-month timeline for establishing procurement vehicles indicates that while the government is moving forward with AI adoption, it is taking a measured approach rather than rushing to deploy tools across the public service.
Support Public-Records Analysis
This analysis is based on government records released under access-to-information laws. If this breakdown was useful, you can support future Government Files work with a one-time tip.
Support Government FilesAll documents referenced are from Invest in Canada, request number A2025-0105950, obtained through Access to Information requests. The records contain a government-wide communiqué from Shared Services Canada dated January 22, 2025, regarding the use of generative artificial intelligence tools across federal departments.
