Bernie Sanders introduced the American AI Sovereign Wealth Fund Act in early June 2026. It's a one-time 50% tax on OpenAI, Anthropic, xAI, and other leading AI companies -- paid in stock, not cash. The shares would be placed in a public sovereign wealth fund; Americans would get voting rights and equal board representation through the fund, plus direct payments and proceeds for healthcare, education, and housing. In parallel, Trump has been in active discussions with OpenAI for over a year about a US government equity stake. On June 6, Trump said the government may take direct equity stakes in OpenAI and xAI. Under that mechanism, OpenAI would donate shares to the federal government to seed what the company has branded a "Public Wealth Fund" (from OpenAI's own April 2026 policy paper).

1. The Public Should Own a Share (Sanders, OpenAI's own framing, populist coalition)

AI was built on humanity's collective knowledge. The wealth concentration coming is unprecedented. Norway did this with oil.

Sanders's argument runs through the source-of-wealth question. AI models were trained on scraped human writing, DARPA-funded research, and university work that taxpayers underwrote. The wealth concentration as AI scales will be enormous. The mechanism: a one-time 50% equity transfer to a sovereign fund, voting rights, board seats, dividend distribution. Common Dreams, NationofChange, and Fortune coverage emphasizes that "ordinary Americans own this" is the political point.

OpenAI itself proposed the Public Wealth Fund structure in April. The company prefers an equity-donation structure over higher corporate taxes -- which is itself revealing about how the AI majors are thinking about the political settlement. From this side, the Sanders bill and the Trump-OpenAI talks are converging on the same instinct: equity transfer is the politically durable mechanism for AI rents to flow back to citizens. Norway's sovereign wealth fund and Alaska's Permanent Fund are the precedents.

2. They Already Pay Taxes (Reason, Cato, market-distortion critique)

Corporate income tax IS the public's share. A 50% equity confiscation is something else, and the regulator-as-shareholder problem is real.

The corporate income tax is the existing public-share mechanism. OpenAI, Anthropic, and xAI all pay taxes; if the public claim is on AI rents, raising the corporate tax rate is the obvious answer. Layering a 50% equity transfer on top of that isn't "the public getting its share"; it's a separate confiscation mechanism with very different incentives and constitutional questions. Reason: Sanders's bill "shows that he doesn't understand AI or wealth." Cato: the structural problem is that "referees with a financial stake in the outcome are no longer calling the game."

Norway is the most-cited precedent and it cuts the other way. Norway's fund was built from oil revenues the government already owned; not from forced equity transfer from private firms. Norway's fund explicitly caps individual company ownership at 10%; Sanders's 50% is unprecedented. From this side, the precedent doesn't support the proposal; it argues against it. Plus the political-control risk: every administration would have direct boardroom influence over which AI is built and for what purposes. The Iran sovereign-wealth-funding-Hezbollah analogy is one of the more pointed critiques.

3. The Real Question Is AI Unemployment (structural)

Whether the public owns 50% of OpenAI or 0% of OpenAI doesn't solve what's actually coming. The income-and-employment problem is the problem.

The equity-stake debate is the politically attractive surface of a deeper structural question. AI-driven labor displacement is the actual policy challenge: knowledge-work automation at speeds and scales the workforce hasn't seen since the agricultural-to-industrial transition. Whether the federal government owns 50% of OpenAI does not determine whether a paralegal, a copywriter, a junior engineer, or a customer-service rep has a job in 2030. The income question is the question.

Mechanisms that actually address displacement run separately. Universal Basic Income, Universal Basic Capital (one-time equity grants to all Americans), wage subsidies tied to AI deployment, retraining funds bundled into AI procurement contracts -- these address the income-and-employment side directly. The Sanders fund proposal would generate revenue for direct payments, which gets partway there; but the equity-transfer architecture is doing political work (Americans "own" AI) that may matter less than whether the income mechanism actually reaches the displaced.

Where This Lands

The pro read says AI was built on collective knowledge and the wealth must flow back to citizens through ownership. But these companies already pay taxes; a 50% equity transfer wrecks regulatory neutrality, and Norway's fund -- the most cited precedent -- explicitly caps single-company holdings at 10% rather than 50%. Stepping back, however, equity ownership doesn't solve what's actually coming -- AI labor displacement at scale.

Sources