Digital public infrastructure to steward and protect the commons as AI concentrates power
AI is generating enormous economic surplus, and left alone, both the surplus and the control over it concentrate in a few hands. Funding the Commons Frontier Fund backs the people building the alternative: funding, governance, and coordination systems that communities, institutions, and autonomous agents can use in the open, that stay auditable and forkable, and that no single company or state can quietly switch off.
This season we're concentrating on the AI-native layer of that work. Programmable funding and treasury systems, coordination and payment infrastructure for autonomous agents, governance that constrains those agents transparently, sovereign and open-source AI, and tools that keep communities in control of their own data and compute. Funding, governance, and coordination mechanisms more broadly are still welcome.
We fund things that are deployed and making impact in the wild. Some projects within this fund have been incubated by Funding the Commons, through our residencies, hackathons, and other activities. Projects may be invited to go on to work with Commons Lab, the new for-profit for-purpose incubator of Funding the Commons, for hands-on support and introductions to institutions that can deploy their work. We back the work in public and publish what holds and what breaks, so the field can build on it.
domain:ai-technologydomain:open-infrastructure
Eligibility
This season's focus: Three lanes, each a version of the same problem: keeping funding, governance, and coordination legible and hard to capture as AI reshapes them. - Public-sector and institutional infrastructure. Programmable treasury and disbursement, outcomes-based financing, verified-outcome-to-payment for climate, health, and livelihoods, last-mile distribution from institutional funders to local recipients, and cryptographic compliance that verifies without centralizing personal data. - Governance infrastructure for AI itself. Agent identity and provenance, agent-to-agent payment and coordination, contribution attribution across multi-agent work, accountability and dispute resolution for autonomous agents, model auditing for AI in high-stakes settings, and sustainability mechanisms for open source in the age of AI-generated code. - Civic and community coordination. Participation and allocation tools that lower the cost of collective decisions, community treasuries and mutual-aid mechanisms, portable reputation and local credit, and data sovereignty that lets communities hold and capture the value of their own data. The fuller map of what we think still needs to exist is the Commons Lab coordination stack. The bar, and this season there's only one: Your project must be live and in real use. - Live, with usage we can verify. Deployed and used by real people outside your team. Activity from distinct users, adoption of your open-source code by other projects, or a named community or institution running it in production. Product analytics for active users help, and we weight what we can check ourselves over self-reporting. - Durable. As a rule of thumb, a couple of months of real use. A strong signal can stand in for time: a named partner running it, real traction/volume, or other projects depending on your code. A six-week-old project with a real institutional pilot clears this. A three-month-old project without users does not. - Open. Code, data, models, and findings are public or on a clear path to public. - Built by a team with skin in the game, who mean to see it through. - A credible path to who deploys it next, ideally a named community, protocol, agency, or ecosystem. Not this round This fund backs live software, protocols, and tools. Events, residencies, and media or storytelling projects aren't eligible this season, and Artizen has a number of incredible funds better suited to those. Also out: idea-stage or pre-launch work with nothing running yet, position papers or speculative research with no working output, and closed or proprietary work. Research is welcome when it ships something usable, a dataset, a tool, or a tested method. One submission per creator or team. Put forward the project you'd stake your name on.
