The global debate about AI sovereignty has focused almost entirely on military applications: whose model flies the drone, whose AI processes satellite imagery, whose system guides missile defence. These are urgent questions. They are also the wrong starting point.
The more consequential question is: whose AI runs the government?
Not (just) the military. The government. The systems that collect taxes, administer benefits, manage health records, process immigration applications, and deliver the services that billions of people depend on every day. These systems are being integrated with artificial intelligence right now, in most cases without public debate, without sovereignty frameworks, and without any coherent strategy for how AI deployment at the federal level connects to the provincial, municipal, or treaty-alliance level.
New research from our team at PatternPulse.AI reveals the depth of this dependency. In Canada, Microsoft Azure consumption across the federal government runs at four times that of all other cloud providers combined. Of the companies approved to sell cloud services to the Government of Canada, seven are American. One is Canadian. Seven to one. And Microsoft itself admitted in a French court that it cannot guarantee foreign government data will be safe from US agency access.
This is not just a Canadian problem, it’s a NATO problem, a Commonwealth problem, and an OECD problem. Most Western democracies run their government digital infrastructure on American platforms, subject to American jurisdiction, accessible in principle to American intelligence agencies under the CLOUD Act. The AI layer inherits the sovereignty characteristics of the infrastructure beneath it. If the infrastructure is not sovereign, the AI running on it is not sovereign either.
PatternPulse AI’s full analysis (available to subscribers) develops three original frameworks for assessing and addressing this crisis:
The Sovereign AI Maturity Model plots countries across three dimensions: Infrastructure Sovereignty, Policy Maturity, and Application Depth. The results are stark. France scores high on all three. The United States scores high on capability but is actively dismantling its governance framework. Canada and Ontario sit in the danger zone, deepening AI integration on foreign infrastructure with minimal policy governance.
The Government Deployment Vertical identifies the critical gap that no existing framework addresses: the absence of any coherent model connecting federal, provincial, municipal, and treaty-alliance AI deployments. Data flows between these levels. A sovereignty framework that covers the federal level but ignores the municipality processing your building permit is a framework with a hole in the floor.
The Sovereign AI Scorecard is a nine-indicator executive assessment tool that any government department can use to evaluate its AI sovereignty posture in an afternoon. Indicators span infrastructure, governance, and operational resilience, including a 90-day operational autonomy test and coherence monitoring based on Evans’ Law. A score below 25 out of 45 indicates a Sovereignty Crisis. Most NATO allies would not pass.
The analysis positions against and extends the major frameworks from McKinsey, the Tony Blair Institute, and the Bennett Institute for Public Policy, identifying blind spots in each, particularly the “sovereignty leak” between levels of government that current models ignore.
The full paper, maturity model, scorecard, and five-year executive roadmap for Tier 2 nations are available through the PatternPulse AI research library.
Pattern Pulse AI provides strategic intelligence on AI architecture, reliability, and sovereignty for enterprise and government clients. For subscription inquiries, contact jen@patternpulse.ai.
