The $30 million toe in the water

Canadian scholars David A. Johnston and Hjalmar K. Turesson from York University's Schulich School of Business propose adopting open-source AI models and domestic infrastructure to secure digital sovereignty, as U.S.-controlled large language models like Anthropic's Mythos raise national security concerns. According to the report, the fundamental issue remains: leading AI models like Mythos are American technologies, giving U.S. firms significant control over IT infrastructure used by Western democracies.

While this is a positive step, the fundamental issue remains: leading AI models like Mythos are American technologies, giving U.S. firms significant control over IT infrastructure used by Western democracies. Canadian scholars David A. Johnston and Hjalmar K. Turesson from York University's Schulich School of Business propose two guiding principles to enhance Canada's digital sovereignty.

An echo of Sydney's 2024 institutional buy-up

Recently, advances in AI have lowered the barriers to developing capable open-source models.. For instance,the Chinese startup DeepSeek released a low-cost, open-source model in January 2025, and open-source alternatives typically follow commercial releases by three to four months with comparable performance. An open-source model rivaling Mythos is likely within six months. This means Canada can build and control its own high-performance AI systems without the massive investments required by models like ChatGPT or Claude, thereby protecting digital sovereignty.

Digital sovereignty involves a nation's control over its data, data processing, and the knowledge generated.. This safeguards personal privacy, intellectual property, and critical infrastructure. Canadians expect their medical records, educational data, and transportation networks to be secure from foreign legal overreach and commercial exploitation.

Who is the unnamed buyer?

According to the report, some universities already run independent AI platforms using open-source models. Moreover, AI efficiency is rapidly improving; for example, Google's Gemma 4 E4B model runs on a smartphone yet surpasses earlier GPT-4 capabilities. efficiency doubles every three to four months, reducing hardware and energy needs.. canada leads the G7 in educational attainment, positioning it well to develop homegrown AI expertise.

By adopting open-source models and domestic infrastructure, Canada can ensure its AI ecosystem aligns with national values and security requirements, reducing reliance on foreign technology and enhancing control over critical digital assets.

What auditors flagged in the May filing

The U.S. CLOUD Act allows American authorities to access data on these servers regardless of location, posing a national security risk. Canadian scholars David A. Johnston and Hjalmar K. Turesson from York University's Schulich School of Business propose two guiding principles to enhance Canada's digital sovereignty.

First, Canadian public and private sector organizations should adopt and adapt open-source AI models for domestic use. Second, the physical infrastructure to run and store these models should be located in Canada and owned by Canadians. Currently, commercial large language models used in Canada are primarily developed by three major U.S. corporations.