Much of the public conversation about AI today focuses on what it can do. New capabilities emerge almost weekly, and with them come understandable questions about safety, trust, and control. But the questions that matter most are: Who controls the infrastructure behind AI? And what values is it designed to protect?
Answering those questions will shape not only how AI works, but whose interests it serves. As governments, businesses, and citizens grapple with the future of AI, we have a critical opportunity today to ensure the building blocks of AI are pro-human by design. The choices we make now will determine whether AI continues to be something owned and directed by a small number of actors — or a resource that can be shaped and governed more broadly in the public interest.
This became even more apparent recently, when the U.S. government’s action to suspend access to Mythos unnerved governments and companies around the world, raising concerns about one government’s ability to unilaterally cut off technology used by others.
Encouragingly, a new path is beginning to emerge, led by middle-power nations. At this year’s World Economic Forum in Davos, Canadian Prime Minister Mark Carney set out a challenge for middle-power nations like his own — nations with the capacity to build a world that “encompasses our values, such as respect for human rights, sustainable development, solidarity, sovereignty, and territorial integrity.” Canada has also released its national AI strategy, which prioritizes the development of open-source AI technology.
More recently, the European Union joined Canada in “placing open source at the centre of the EU’s technological sovereignty.” This includes commitments to support open-source alternatives throughout the AI stack, backing startups, and developing new government procurement guidelines that put the EU’s thumb on the scale in favor of open-source innovation. Countries from Germany to Japan are looking at ways to integrate open source into their national strategies, while the UK has announced an Open Source Builder’s Fund, aiming to make Britain the “home of global open source AI talent.”
Open source is emerging as a powerful consensus middle path and the private sector is catching on.
Research has shown that open-source technology has created over $8.8 trillion in demand-side value: firms would have to spend 3.5 times more on software than they currently do if open source didn’t exist. Entrepreneurs and researchers are building and leveraging open-source AI tools, models, and datasets that reflect local needs and perspectives. Developers are gravitating quickly toward open-source AI. A recent a16z and OpenRouter study found that open-source models grew from roughly 1–2% of token volume in late 2024 to nearly 30% by mid-2025 — a clear marker of momentum among builders, and a significant business opportunity.
Unlike AI technology owned and controlled by a few large corporations, open-source AI is available to everyone. That not only means governments and companies can own the infrastructure on which they build — it means anyone can look under the hood. That transparency is what can make AI safe and accountable by design.
At Mozilla, we often talk about building technology that amplifies human agency rather than replacing it. In December 2022, we amended our foundational Manifesto with a Pledge for a Healthy Internet, centered on four commitments: that the internet should include all people; promote civil discourse, human dignity, and individual expression; elevate critical thinking and verifiable facts; and catalyze collaboration across communities working for the common good.
That is the path Canada, the EU, and other middle powers are leading — in collaboration with a private sector hungry for AI alternatives to the closed models that dominate today.
AI has enormous potential, but a growing unease surrounds its direction and who controls it. The antidote is building AI that is open, trustworthy, and reflective of a diversity of voices outside Silicon Valley and China’s AI labs. That future can only be built by a middle-power, open-source coalition determined to ensure AI works for human beings — not the other way around.
The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.
