How to Achieve Independence, Privacy, and Trust in the Adoption of AI

In my week at Open Source Summit Europe and AI_Dev in Amsterdam, the topic of digital sovereignty persisted throughout various keynotes, panels, and hallway track conversations. Control, agency, and participation are seen as critical for Europe’s digital future. But sovereignty does not necessarily equate to solutions built within a country’s borders. Instead, it is seen as a larger movement of capacity-building that places local developers and innovators as builders and decision-makers on the global open source technologies the country or region relies on. This is as relevant in Europe as it is in my home country of Canada, where concerns around digital sovereignty also abound. 

Governments and businesses are at an inflection point in the development of AI, with the question of sovereignty also looming large in this context: How can we build AI infrastructure for us—AI in our own languages, using our own data, reflecting our own norms and context? 

Earlier this year, LF Research, Futurewei, and LF AI & Data started a project with the aim of evaluating whether “Sovereign AI” was a strategic priority or simply a fad, where in the world it was felt most, and how open source and global collaboration may play a role in this movement—despite its orientation as a localized, siloed concept. 

We ran a survey in our community, collecting data from over 200 respondents, and hosted webinars with experts in three different geographies to capture global perspectives on the topic. Our data shows just how important Sovereign AI has become, and why. The following describes a few key findings from our research. 

Why Sovereign AI

So, why is the practice of sovereignty important to our community? The study highlights a few big drivers. First, control over data is a top concern: Over 70% of respondents want to keep sensitive data safe from outside appropriation and avoid using black-box solutions without transparency into the storage and safety of data. This also speaks to a concern for national security, and removing the risks that come with dependence on foreign corporations. Another key driver is the ability to localize a model within the specific context of the organization or country, embedding regulations, cultural norms, and local languages that reflect the user’s reality back to them as accurately as possible. 

The role of open source

Despite the localization and protectionism of Sovereign AI, we collected reassuring findings on the value of openness for this practice. In fact, the foundation of Sovereign AI includes pen source software, open standards, and open data.

Open source received overwhelming support in our survey, with nine out of ten people calling it essential. Reasons for this included open source’s transparency, security, flexibility to customize, and faster innovation through collaboration.

The study shows that open source and Sovereign AI really work hand in hand: 94% of respondents said global collaboration is essential for sovereign AI. The biggest areas for collaboration were foundation models and datasets (both at 59%), followed by development tools and platforms (39%).

Roadblocks ahead

Of course, challenges and roadblocks exist. Respondents pointed to a range of obstacles, from poor data quality and availability, a lack of technical expertise, geopolitical tensions and national security restrictions, and regulatory limitations. While the idea of sovereign AI is gaining traction, turning it into a sustainable reality will take investment, talent, and smart governance.

Governing Sovereign AI

Another key finding is that its governance will likely be shared. National governments will play a significant role, but so will open source foundations and community-driven projects. This hybrid approach reflects the balancing act of local control alongside global participation, a key theme found throughout the study.

Recommendations for its governance are clear. Invest in open source AI infrastructure, train and upskill homegrown talent, support open source foundations and strategic partnerships, and improve data quality and accessibility. 

This is a critical moment in AI development. Sovereignty is a strategic priority for governments and organizations around the world, and open source is the key enabler. Through the use of, and contribution to, and participation in open source projects, nations and organizations can retain control while benefiting from collective progress. The report lists some key resources for development and training on this topic—take a look and get involved! 

Anna Hermansen is a senior researcher at the Linux Foundation.