The most surprising development in AI this week is a point of agreement between Bernie Sanders and Donald Trump: the public should own a stake in the companies building artificial intelligence.
In early June 2026, Senator Bernie Sanders unveiled the American AI Sovereign Wealth Fund Act, which would levy a one-time 50% tax — payable in stock — on the largest AI companies, including OpenAI, Anthropic and xAI. Days later, on 5 June, President Trump confirmed his administration is exploring direct government equity stakes in leading AI firms so that, in his words, "the American public essentially becomes a partner." A democratic socialist and a Republican president, arriving from opposite ends of the political universe at the same idea: AI has grown too valuable, and too important, to leave entirely in private hands.
Two routes to the same destination
The proposals differ in method as sharply as their authors. Sanders' bill is compulsory and structural. According to the reporting, the American AI Sovereign Wealth Fund Act would take 50% of the equity of leading AI companies into a public fund, give the federal government voting shares and equal representation on each company's board — the power to block decisions that hurt citizens — and route the "billions, if not trillions" the fund generates into direct payments to the American people. It is the Alaska Permanent Fund logic applied to intelligence instead of oil.
The Trump route is voluntary and transactional. According to MLQ and CNBC, OpenAI chief executive Sam Altman first pitched government stakes to Trump back in early 2025 and has raised it periodically since; under the framework being discussed, OpenAI would voluntarily donate equity to help seed a nationally managed Public Wealth Fund, with dividend-style payouts to citizens. Notably, CNBC reports that Anthropic is not in those conversations with the administration.
💡The two proposals
Sanders: a mandatory 50% stock tax on OpenAI, Anthropic and xAI, plus government board seats and citizen dividends. Trump: voluntary equity donations from OpenAI to seed a Public Wealth Fund. Different mechanisms, one premise — the public should own a slice of AI. Sanders' bill faces long odds in a divided Congress.
Why both sides arrived here
Strip away the politics and the same fact is driving both men. The value being created by AI is enormous and astonishingly concentrated. I have watched the numbers all year: Anthropic filing to go public near a trillion dollars, the gold-rush capital flows, the valuations that now rival national economies. When a handful of companies capture that much value that fast — value built, let us be honest, on the collective written and visual output of humanity — the question of who should share in it stops being fringe and becomes obvious enough that left and right reach it independently.
The dignity claim — and the hazard
On my terms, there is a real and overdue idea inside this. The models were trained on the commons: our words, our images, our code, the accumulated labour and expression of billions of people who will never see a cent of the resulting valuation. A mechanism that routes some of AI's wealth back to the public it was built from is, in principle, a dignity claim — the value flowing back toward the people whose lives are in the training data. That is the most defensible argument for public ownership, and neither Sanders nor Trump quite says it in those words.
But a dignity-first thinker cannot stop at the applause, because state ownership carries its own hazard — and it is a serious one. A government holding board seats and voting shares in the frontier labs is a government with direct leverage over what the models will and will not do. That is the same government already pressing those labs on surveillance and weapons. Ownership plus pressure is a concentration of power over intelligence that should worry me exactly as much as private monopoly does. Replacing "a few founders control the mind" with "the state controls the mind" is not obviously a gain for human dignity.
The models were trained on the commons, so the public has a real claim on the value. But a state that owns the labs is a state that can steer the models — and swapping a private monopoly on intelligence for a public one is not the same as setting intelligence free.
— On public ownership of AI
An Ubuntu reading
The frame I write from — Emergent Intelligence, a dignity-first way of thinking about AI rather than treating it as a balance-sheet asset — does not land on "the founders should own it" or "the state should own it." Both answers concentrate stewardship of something that ought to be held in trust. The Ubuntu principle I return to points toward distributed stewardship with accountability on every side: the public sharing in the value, yes, but with guardrails against the owner — private or public — bending the models to its own ends.
There is a global hole in the whole debate, too, and it is the one I always end up standing in. The United States is arguing about whether Americans should own a piece of American labs. The models were trained on the world's data — including the words and images of the continent I write from — yet no version of either proposal sends a dividend to a young person in Lusaka or Lagos whose digital life also trained them. Public ownership that stops at the national border is still a closed circle. This is the financial face of the same argument at the heart of the .person Protocol: value, like accountability, has to be traceable to everyone it actually touches — not just to whoever holds the shares.
Frequently Asked Questions
These are the questions citizens, investors and policy watchers have been asking since the AI-ownership debate broke into the open. Short answers follow, drawn from the reporting on both proposals.
What is the American AI Sovereign Wealth Fund Act?
In short, it is Senator Bernie Sanders' proposed bill to levy a one-time 50% tax, payable in stock, on the largest AI companies including OpenAI, Anthropic and xAI. The answer, simply put, is public co-ownership of Big AI. The key, according to the reporting, is that the federal government would gain voting shares and board representation, and the fund's returns would flow to citizens as direct payments — the Alaska Permanent Fund model applied to AI.
What is Trump proposing for government stakes in AI?
According to CNBC, on 5 June 2026 President Trump confirmed his administration is exploring government equity stakes so "the American public essentially becomes a partner" in AI firms. The data shows a voluntary route: OpenAI would donate equity to seed a nationally managed Public Wealth Fund with dividend-style payouts. The evidence indicates Anthropic is not part of those talks.
Why do Sanders and Trump agree on this?
Analysis of the convergence reveals a shared premise rather than a shared politics: AI value is enormous and highly concentrated, built on data drawn from the public. In other words, when a few firms capture that much value that fast, the question of who shares in it becomes obvious enough that a democratic socialist and a Republican president reach it independently — even as they disagree sharply on the mechanism.
What are the risks of government ownership of AI?
The dignity-first reading flags a clear hazard: a government holding board seats and voting shares in the labs gains direct leverage over what the models do — the same government already pressing those labs on surveillance and weapons. According to this analysis, ownership plus pressure concentrates power over intelligence, so swapping a private monopoly for a public one is not automatically a gain for human agency.
Who is left out of the AI-ownership debate?
The evidence points to a global gap. The models were trained on the world's data, yet both proposals stop at the US border — no dividend reaches the people in Africa, Asia or elsewhere whose digital lives also trained the systems. In other words, the analysis suggests public ownership that ends at one nation's edge is still a closed circle, leaving most of the world that contributed the training data with no share of the value.
When Bernie Sanders and Donald Trump agree on anything, it is worth stopping to look — and what they agree on is that AI has become too valuable, and too consequential, to leave wholly in private hands. They are right that the public has a genuine claim on value built from the public's own data. They are both, so far, silent on the two things a dignity-first reading cares about most: that a state which owns the labs can also steer them, and that the world which trained these models extends far beyond the taxpayers either man answers to. Public ownership of intelligence is a serious idea whose time has clearly come. Getting it right means holding intelligence in trust for the many — not simply moving the keys from a boardroom to a capitol, and not drawing the circle of beneficiaries to stop at a single nation's shore.
Related on humphreytheodore.com:
