Give a clear mandate.
People decide the goal and limits. The agents cannot invent a new mission or award themselves more power.
An idea worth building together · Proposal v0.1
AI that answers to us.
The people it serves should decide what powerful AI is for—and what it is allowed to do.
Commonwealth Intelligence proposes a voluntary organization where AI’s purpose and authority come only from the people it serves. Through agreed rules that protect everyone’s rights, they would set its limits and could change or withdraw its authority. Founders, funders, and AI suppliers would be bound by those same rules.
An early proposal, not an institution already speaking for anyone. Support and comment signup are not open yet.
01 / Why care?
If AI helps shape our lives, we should have a say.
Useful AI is only part of the question. Who gets to tell it what to do?
This proposal puts that authority with the people it serves, through equal standing, protected rights, and rules that bind everyone—including the founder. A company could supply technology without getting to set the community’s purpose.
People would choose the priorities, question decisions, and replace tools or leaders that fail them. No AI could assume your consent, silence disagreement, or take control of money or power.
We should shape this while we still have room to choose.
As AI becomes part of everyday life, changing the systems we depend on may become harder and more costly. We should start building meaningful public control now.
We do not need every answer before we begin. We need a shared starting principle: the people AI serves should give it its purpose, set its limits, and retain the power to change or replace it.
02 / What could that look like?
People choose the goal. Agents help do the work.
AI agents are software that can carry out a series of tasks. They are one way to put the idea to work.
Imagine a community deciding to help people learn new skills. Under its agreed rules, it approves a grant program. The community authorizes AI agents to explain the rules, help with paperwork, and track requests—not to change eligibility, decide awards, or release money. Applicants check their own paperwork. People authorize the grants and remain responsible for the results.
Keep a way to challenge it.
You can question an answer and ask for human review. Rights and fair appeals do not depend on agreeing with the majority.
Change what fails.
Through the agreed process, the community can change or withdraw the agents’ job. Humans remain responsible for pending requests and appeals. Faster paperwork alone would not prove the system is fair.
A fictional example, not a grant or agent service available today. The starting proposal is a lawful, voluntary organization; it claims no power over people outside it.
03 / Help shape the idea
Would you support AI governed by the people it serves?
A support opt-in would be like signing a petition: “I want this idea developed and tested.” You would not be approving every detail, joining a government, or giving anyone authority to speak for you.
You could support the goal and still disagree with the proposed design. Support would be confirmed by email, and you could withdraw it.
Have a comment instead—or as well?
A question, a technical concern, a philosophical objection, or a practical suggestion: all would be welcome. You would not need to pledge support to offer a critique.
Comment intake is not open yet. Nothing you write is collected here. See the questions we are working through →
04 / Look closer if you want
The goal is simple. The design needs scrutiny.
Who takes part? How are choices made? What prevents capture or abuse? Proposal v0.1 offers starting answers, with tradeoffs and open questions.
Read the proposed safeguards and worked examples
Protected rights, equal civic standing, limited powers, independent oversight, fair appeals, and no permanent founder veto. None of these rules has been adopted. High-stakes Internet voting remains an unresolved security problem.
Who is behind Commonwealth Intelligence?
Jeff Lerner started the proposal, with funding and development help from the 4th P. It began with a founder, not a community vote, and does not yet represent constituents. Founders and funders would be bound by the same rules as everyone else.
Disagreement is welcome. Read the risks or explore the open questions.
Support-list status and what a pledge would mean
Confirmed supporters
Registry not activatedSupport signup is closed. No one is being counted.Count only confirmed support that has not been withdrawn. An email confirmation checks an inbox, not unique personhood. A supporter is not a verified constituent or voter. Never publish email addresses.