AI is built on all of us. Almost none of us own it. This page is the argument for changing that, in six steps, each one made properly in a post you can read. It is the question this site exists to answer, and it stays open until the world settles it.

1. A handful of people hold the votes. Five companies build the machines the world now runs on, and five men hold the votes that decide what they do. Not the money, the votes. When Britain electrified, a Conservative government put the grid under public control within a decade, because nobody thought a handful of owners should decide who got light. Nobody has thought that about AI yet.

2. Everyone already pays for it. Meta's flagship data centre mostly belongs to a private lender, financed with pension money: ordinary savers bankroll the machine and collect interest, and get no say. One in six British workers pays for AI tools out of their own pocket and hands the time saved back to the boss. And the archives the models learned from were sold by institutions, with nothing for the people whose work fills them. The money goes in from everyone. The control stays with the few.

3. Regulation makes the owners behave. It does not change who the owners are. Regulation civilises the arrangement: the owners keep the machine and a court makes them serve everyone. That is the polite outcome, and it is what every government is reaching for. It leaves the concentration exactly where it found it.

4. A payment is not a vote. Every plan to share AI's wealth moves proceeds and none moves control. A dividend from a fund somebody else runs is a cheque, not a seat. And the one company that says humanity owns it proves the point: the OpenAI Foundation holds every vote and a quarter of the money, and its board picks itself. Humanity got all the control and none of the choosing.

5. The companies open places to talk, never places to decide. In one week in September, Microsoft opened a consultation, OpenAI a foundation and Google DeepMind an institute. Three invitations to speak. Not one where anyone outside the company could stop anything.

6. So: owned by the people who use it. Not nationalised, not regulated harder, owned. The idea pages work out how. An AI owned by the people who use it: a locked society, a product that carries the membership, a brake rather than a steering wheel. The user-ownership trust. Humanist means owned by humans. Build things together, and keep them. Eight ideas a few thousand people could start next month. None of it needs a government to go first.

What would prove this page wrong. The bets are on the scoreboard, dated. That no G7 government takes a controlling stake in a frontier lab by 2029. That the FCA's register shows a new mutual built to own an AI by the end of 2027. That a benefit-sharing deal gives a poor country a vote rather than a payment by the end of 2027. If ownership spreads without anyone building the mutual, this page was wrong about how it happens. If the transfers flow and the votes never move, it was right, and the cheques were the anaesthetic.

Start with one. If you read nothing else, read The Machine Has No Members. It is the shortest version of the whole argument: no AI company has members, only customers, and everything else follows from that.