Mozilla president says open-source AI push is gaining ground
Mark Surman argues Canada, the EU and other governments are using open source to reduce reliance on closed AI systems.
By Hana Yoshida · Markets Reporter
3 min read
Mozilla President Mark Surman says governments and developers are moving toward open-source AI as a counterweight to systems controlled by a small group of large technology companies. In a Fortune commentary, Surman argued that the issue will shape who controls AI infrastructure and whose interests the technology serves.
Surman, who leads Mozilla’s nonprofit work and its portfolio of public-benefit technology efforts, framed the debate around control, safety and public accountability. He said concerns intensified after the U.S. government moved to suspend access to Mythos, an action he said unsettled companies and governments by showing how one government could cut off technology used elsewhere.
According to Surman, “middle-power” governments are beginning to back open-source AI as a way to preserve more control over critical technology. He pointed to Canada, the European Union, Germany, Japan and the United Kingdom as examples of governments weighing or adopting open-source policies in AI.
Canada and Europe take prominent roles
Surman cited Canadian Prime Minister Mark Carney’s remarks at the World Economic Forum in Davos, where Carney called on middle-power countries to help build a global order aligned with values including human rights, sustainability, solidarity, sovereignty and territorial integrity. Surman also noted that Canada’s national AI strategy gives priority to open-source AI development.
The European Union has also put open source at the center of its technology-sovereignty agenda, Surman wrote. He said the EU’s plans include support for open-source alternatives across the AI stack, aid for startups and government purchasing rules designed to favor open-source innovation.
Surman said other governments are following related paths. Germany and Japan are looking at how open source fits into national AI strategies, while the U.K. has announced an Open Source Builder’s Fund intended to attract open-source AI talent.
Developers and companies add momentum
Surman argued that open source is also gaining support from the private sector because it lowers costs and gives builders more control. He cited research from Harvard Business School finding that open-source technology has generated more than $8.8 trillion in demand-side value, and that firms would need to spend 3.5 times as much on software if open source did not exist.
He also pointed to a study by a16z and OpenRouter that found open-source models rose from about 1% to 2% of token volume in late 2024 to nearly 30% by mid-2025. Surman described that shift as evidence that developers are adopting open-source AI tools, models and datasets quickly.
Surman contrasted open-source AI with closed systems run by large corporations, saying open systems let governments and businesses own more of the infrastructure they depend on. He also argued that transparency allows outsiders to inspect AI systems, which he said can support safety and accountability.
Mozilla’s own approach, Surman wrote, is rooted in building technology that expands human agency. He cited Mozilla’s December 2022 update to its manifesto, which committed the organization to an internet that includes all people, supports civil discourse and human dignity, elevates verifiable facts and encourages collaboration for the public good.
Surman said open-source AI could give governments and companies alternatives to the closed models that dominate the field today. He argued that a coalition of middle powers and private-sector builders could help ensure AI reflects a wider range of voices beyond Silicon Valley and China’s AI labs.
This story draws on original reporting from Fortune.