The bills, agencies, and rules shaping AI in Canada, from Bill C-16 to the CRTC and the Privacy Commissioner. The chapters on who sets the guardrails.
The previous chapters have repeatedly returned to one finding: on multiple AI policy dimensions, Canada in 2026 has weaker regulatory protections than peer jurisdictions, with the gaps concentrated in specific identifiable places — political deepfakes, comprehensive privacy frameworks for AI, mandatory data-centre disclosure, training-data consent requirements, workplace AI surveillance limits, Indigenous data sovereignty integration. Each chapter has noted the gap relevant to its specific topic. This chapter steps back and asks the institutional question: how is AI actually governed in Canada, and why does the regulatory landscape look the way it does?
The answer is that Canada's AI governance is fragmented across federal departments, provincial jurisdictions, sectoral regulators, professional bodies, and international commitments, without a comprehensive coordinating framework, without a designated AI regulator, and without the binding legislative authority that comparable jurisdictions have established. The federal AI for All strategy is investment-focused rather than regulatory-focused. The Artificial Intelligence and Data Act (AIDA) that would have established comprehensive federal AI regulation died with Parliament's January 2025 prorogation, and the Carney government has not committed to reviving it. The result is the institutional environment the rest of the guide's analysis has been operating against.
One federal bill is moving, though it is online-harms legislation rather than AI law. In June 2026 the Carney government tabled the Digital Safety Act, Ottawa's fourth attempt at online-harms regulation since 2021. Its headline is a ban on social-media accounts for Canadians under 16, with platforms able to apply for an exemption if they can show adequate safeguards; it would also create a Digital Safety Commission of Canada, a standing federal digital regulator of a kind the AI file has so far done without. It is worth watching for two reasons. First, it is the clearest current test of whether Ottawa can stand up a digital regulator with real authority rather than another voluntary code. Second, it sits in the familiar gap between announced and operational: officials concede the under-16 ban would realistically not apply until late 2027 or 2028, once the bill passes and the commission is built. The AI connection is indirect but real, because the harms it targets (recommender systems tuned for engagement, the data they harvest on minors) are AI systems, and a commission built for digital safety is the kind of body a future AI mandate could later be handed to. Three earlier attempts died before passing; whether this one survives the same competing pressures is the open question.
This is not the same as saying Canada has no AI governance. Multiple institutional resources are working on AI policy: the Office of the Privacy Commissioner, provincial privacy commissioners, the Treasury Board (within the federal public service), sectoral regulators (CRTC, OSFI, Health Canada, others), the Auditor General, the judiciary. The institutional resources are real. They are also working without the comprehensive coordinating framework that AI policy at the current scale of deployment would typically require, and the guide asks readers to track both facts simultaneously.
Start with the decision you want someone to answer for. A federal department using a tool, a province setting privacy rules and a regulator overseeing an industry each have a different job. The failed AIDA proposal, the treaties and the examples from abroad help explain how we got this mix and which institutions could change it.