Mila and Mozilla announce new initiative to build trustworthy open source AI for everyone, with Canadian government support

Today at ALL IN, Canada’s largest AI and technology event, Mozilla and Mila announced a new initiative and fresh investment to build an open source AI foundation layer that enables organizations and institutions to own and operate advanced AI systems locally, ensuring full control over their technology and data. 

Doubling down on its commitment to open source AI, the Government of Canada announced its support for the initiative. Mila will lead the technical delivery and coordination of the project, while Mozilla contributes technical expertise. It also provided the initial $5 million investment to kick off the project. Hypertec is committing an additional $1 million in first-year funding to accelerate initial Canadian deployments of the open source AI foundation layer on Hypertec hardware. 

“Canada has a choice: depend on technologies developed elsewhere, or build more of what we need here at home,” said The Honourable Evan Solomon, Minister of Artificial Intelligence and Digital Innovation and Minister responsible for the Federal Economic Development Agency for Southern Ontario. “Open source AI gives Canadian businesses and institutions greater control over their technology and data, while making powerful tools more affordable, accessible, and easier to adapt. By supporting this work, we are strengthening Canada’s capacity to build and adopt AI on our own terms.” 

Mila and Mozilla will lead the work together: Mila drawing on a world-class community of close to 2,000 researchers and professionals, Mozilla as technical partner from industry, bringing 25 years of experience stewarding open infrastructure others build on. Mila and Mozilla are actively engaging new partners, inviting companies, research institutions, funders, governments, and developers to join and support this work. 

What’s being built

The goal is simple: make owning your AI as easy as renting it. Using open source models was never the hard part. Turning a raw open source model into something a small business, a hospital, a local charity, or a government can run in production — secure, reliable, plugged into everything else — takes an engineering team most organizations and institutions do not have and months they can’t spare. 

The goal is to offer businesses and organizations a ready-to-use AI package that they can run privately and keep under their own control, rather than having to build a complex system themselves or rely entirely on expensive, pay-per-use proprietary AI services.

For example, a small manufacturer could use the system as a private AI assistant for its employees. It could search the company’s manuals, procedures, and past project files; help staff draft reports or answer technical questions; and help its software team write and improve code. Because the system is designed to be able to run locally, the company could do this while keeping its proprietary information and data within its own environment.

That also means lower costs. For the vast majority of everyday business tasks, companies will be able to use open source AI running on open source tools instead of paying a commercial provider every time an employee makes a request. But this requires reducing the technical work and expense involved in putting open source AI into practice.

This initiative was founded to solve that problem, building the free layer that makes owning open source AI easy — the same kind of layer the web was built on. It has two halves: An open standard, published as interface contracts, so any part of the stack can be swapped for a better one. And a working “reference implementation” any organization can install on its own machines or the ones it chooses, running the models and controls it chooses against its own data, with governance and access control built in from the start. The ambition is that with the standards and foundation in place, open source AI can be built anywhere, by anyone. 

“Six months ago, we announced our partnership with Mozilla to advance open source and sovereign AI. Today, we are bringing that work to a whole new level,” said Valérie Pisano, President and CEO of Mila. “By delivering an open, secure AI foundation layer, we empower organizations, from small businesses to non-profits to governments, to own their technological future so they can run, control, and maintain AI models themselves. At Mila, our research community has always believed that for AI to be trustworthy and accessible, it must be built on open standards that keep control in local hands.”

The work already underway

Over the last six months, Mila and Mozilla have been designing the architecture, deciding which open source components belong at each layer, and testing that the whole system works end to end. This investment builds on that progress. 

Working alongside Mila and Mozilla, Hypertec will help move the initiative from research and reference implementation to real-world adoption by Canadian businesses and institutions, providing a practical, private, and cost-effective path to deploy AI while maintaining greater control over their data and technology.

“AI is advancing at an extraordinary pace and has the potential to transform our economy and society for the better. Canada has an important role to play in ensuring that AI is developed and adopted responsibly, said Simon Ahdoot, CEO of Hypertec Group. “Hypertec is proud to help turn open source innovation into AI that Canadian organizations can deploy securely and under their own control. This is exactly the kind of partnership between government, research, and Canadian industry needed to realize the full potential of AI.”

Why now

Mozilla’s State of Open Source AI report found that while 79% of developers adding AI functionality use open models, only 53% of teams ever reach production, stopped by cost, security, integration, and maintenance. This work aims to change that.

“AI today is at a crossroads, where it could be closed and owned by a few, or open and available to every coder, developer, enterprise, and nation,” said Mark Surman, President of Mozilla. That’s what we’re building — an open source AI ecosystem that fits together as seamlessly as the web, and that anyone, anywhere can build on.  We’re so grateful to Canada for scaling this work, and call on partners across sectors — from enterprises and governments to coders and startups — to join us in building this future.” 

Within six months, Mozilla and Mila expect to publish working reference implementations for enterprise, government and public-interest use cases. The two-year ambition is bigger: to have solved this problem outright, so that open source AI can be adopted fully and easily, anywhere, by anyone.