hello@openfuture.eu
What if AI’s most powerful tools were in everyone’s hands, not just Big Tech’s?
On May 20, 2025, we hosted an online event discussing the future of Public AI. We presented our latest white paper—commissioned by Bertelsmann Stiftung—exploring how to make state-of-the-art AI more accessible and accountable.
The white paper, authored by Felix Sieker, Alek Tarkowski, Cailean Osborne, and Lea Gimpel, examines the requirements for AI systems as Public Digital Infrastructure. It analyzes the AI stack, focusing on power concentrations and ways to make AI components more public. While advanced AI systems remain largely under private company control, our research offers a promising alternative: Public AI systems built on transparent governance, public accountability, and fair access to core components.
The webinar featured the paper’s authors, Felix Sieker and Alek Tarkowski, in conversation with industry and policy experts Renata Ávila (Open Knowledge Foundation) and Luca Cominassi and Albert Cañigueral (Barcelona Supercomputing Center).
To start, Alek Tarkowski introduced the event’s focus on AI as public infrastructure, exploring AI systems that serve public purposes and democratic oversight. The timing is particularly relevant in Europe—and beyond—as the EU has announced the AI Continent initiative, approved the AI Act, and launched the Data Union Strategy, which will help inform this policy debate.
Felix Sieker guided the 50+ online attendees through the report, which can be downloaded here, providing an overview of public AI, the research objectives (demystifying AI, identifying bottlenecks and dependencies, and outlining paths to public AI) and the white paper’s two main contributions:
Renata Avila called for a compelling narrative to help taxpayers understand public AI, drawing parallels with public education’s evolution 200 years ago. Just as education transformed from an elite privilege into a universal right, access to public AI is becoming necessary for broader societal participation. History shows us that systems can be built to serve everyone.
Luca Cominassi and Albert Cañigueral emphasized the need for fully open-source models, noting that “they enable an ecosystem to face common challenges.” They advocated for public AI adoption through procurement and competition, suggesting public administrations should be early adopters of public AI. They highlighted interesting correlations between public administration’s transparency requirements and using public data for public interest.
