Public AI Network

The Public AI Network (PAINT) is an international coalition building AI as public infrastructure — provisioned like electricity, water, roads, libraries, or the internet itself — and coordinating the civil society, research, and policy work needed to make that vision actionable. Founded in summer 2023 by three researchers who convened at the Internet Archive's DWeb Camp in California, the network has grown to over 350 members from more than 100 organizations and serves as the primary coordinating body for the global movement to establish publicly-owned and democratically-governed AI systems as a counterweight to the concentrated private AI ecosystem.

Founding

Joshua Tan (Metagov), Nick Garcia (Public Knowledge), and SJ Klein (Concordance Project) began talking about public AI in summer 2023 at DWeb Camp — the Internet Archive's decentralized-web gathering held in the California redwoods — where each had been pursuing complementary threads independently. They reconvened at "AI Palace" later that summer to draft the coalition's foundational whitepaper and launched the Public AI Network's Slack community that fall. The Rockefeller Foundation provided early funding to support coalition-building in under-resourced AI policy areas. B Cavello, Director of Emerging Technologies at Aspen Digital, joined the core organizing team through her board position at Metagov, and Aspen Digital became the coalition's primary convening partner — co-organizing the Public AI Seminar, publishing the coalition's framework reports, and convening more than 150 participants from 100+ organizations across five in-person events in the coalition's first two years.

Coalition scale and structure

By late 2025, the network encompassed over 350 members from more than 100 organizations worldwide, including Metagov, Aspen Digital, Open Future, Mozilla, the Rockefeller Foundation, the Internet Archive, Public Knowledge, Code for Science and Society, Chatham House, the Berkman Klein Center, and the Bertelsmann Foundation. The coalition operates through weekly coordination calls among core members, a Slack community connecting experts across public AI labs and civil society groups, quarterly public community calls, and working groups on access, accountability, and public-goods governance models. A landmark in-person convening came in August 2024 — an all-day launch event co-hosted at the Library of Congress with 79 attendees from over 60 organizations. An early 2025 gathering in Paris, timed around the Paris AI Action Summit, brought the coalition's international dimension into focus.

Libraries program

The coalition's most direct community-engagement initiative is its public libraries program, developed in partnership with the Library of Congress, Metagov, and Public Knowledge, with the Utah State Library and New Jersey State Library as early state-level partners. The program brings AI kiosks — cloud-based workstations providing access to AI chat, creative generation, coding support, and educational resources as shared infrastructure — into public library branches. By late 2025, six libraries in Utah, Georgia, New Jersey, Texas, and Massachusetts were running pilots. The program's long-term vision extends from tool access through community knowledge preservation to democratic governance: libraries are positioned as sites where communities own and govern their own knowledge through locally-governed AI models, not just access corporate platforms via library hardware.

Publications and intellectual framework

The coalition's August 2024 whitepaper (B. Jackson, B Cavello, N. Garcia, J. Tan, S. Klein, and colleagues) established the foundational conceptual framework. Subsequent major publications include Mozilla Foundation's Public AI: Making AI Work For Everyone, By Everyone (September 2024, Nik Marda, Jasmine Sun, Mark Surman) and Aspen Digital's June 2025 Intelligence in the Public Interest, which laid out a community-benchmarks framework for public influence over AI systems. The 2025 "Airbus for AI" proposal (Joshua Tan, Brandon Jackson, Robin Berjon, Diane Coyle) advocated for a European public AI institution modelled on cross-national aerospace cooperation. The Public AI Seminar — co-organized by Joshua Tan, Alex Krasodomski (Chatham House), and B Cavello — has run four seasons as of mid-2026, with speakers including Yoshua Bengio, Diane Coyle, Lawrence Lessig, and Kim Stanley Robinson.

Infrastructure pilots

In parallel with policy work, the coalition supports concrete infrastructure experiments. The Public AI Inference Utility launched in Q3 2025 providing access to open models across multiple compute environments; a Swiss Public Inference Utility operates at national scale. The AI Commons initiative is reconstructing a public knowledge base for AI training data. SJ Klein's Concordance Project is piloting community-led data pools in agriculture — farming data, regional knowledge systems, and land-use practices governed by farmers rather than technology companies. Each pilot is designed to demonstrate that AI systems can be built and governed as genuine commons.

Movement framing

The Public AI Network's central conceptual contribution is framing AI infrastructure as a public utilities question — analogous to the policy choices that produced publicly-owned water systems, public libraries, and public broadcasting. The coalition positions the democratic-governance layer as inseparable from the technical-infrastructure layer: publicly-provisioned AI that replicates private AI's governance structures is not genuinely public. With over 25 publications associated with the movement by 2025 and growing citation in academic and policy literature, the network has made "public AI" a legible policy option in AI governance debates internationally.