The idea is unusual enough that it took a moment to register: the United States government, owning a slice of the companies building frontier AI.
According to a report by NOTUS, senior US officials have held preliminary discussions with major AI companies about exactly that, the federal government acquiring shares in the firms at the centre of the technology it is also trying to regulate.
The thread runs through OpenAI. Sam Altman, the company’s chief executive, who has spent recent weeks in Washington urging Congress to fund AI testing rather than require model approvals, has discussed the stake concept with senior Trump administration officials periodically since the president’s second term began, NOTUS reported, having first pitched it directly to Donald Trump in early 2025 and raised it again with officials in recent weeks.
The discussions, per the report, have centred on firms voluntarily ceding shares to the government rather than the government buying in.
What the government would do with the returns is part of the pitch. One option discussed is directing the proceeds to public purposes, including a dividend paid to all American households, a framing that turns government ownership of AI into a mechanism for distributing the technology’s gains directly to citizens.
It is a politically potent idea at a moment when, by the polling NOTUS cites, 55% of Americans think AI will do more harm than good in their daily lives, and when OpenAI’s own corporate structure is already under a microscope through the Musk v. Altman trial over its nonprofit-to-profit conversion.
That public scepticism is the context the companies are reacting to. Tech firms are looking for ways to win over Americans wary of AI, and a structure that hands the public a stake, literal or financial, in the upside is one answer to the charge that the gains are flowing to a narrow few. Voluntarily ceding shares reads, in that light, as much as reputation management as fiscal policy.
The objection is structural and immediate. A government that owns shares in a company it also regulates is both shareholder and referee, a conflict critics quoted in the report flagged at once. The incentive to protect the value of a holding sits awkwardly beside the duty to police the company that holds it, and no amount of public-dividend framing fully resolves the tension.
The idea also has unlikely company across the political spectrum. Versions of public equity in AI have been floated from the left as well, on the logic that if the technology displaces large numbers of workers, the public should hold a claim on the resulting profits rather than watch them accrue entirely to private shareholders, a concern OpenAI itself has gestured at even as Altman plays down the idea of an AI jobs apocalypse.
That a Trump-administration conversation and progressive proposals arrive at a similar mechanism, the state holding a stake on the public’s behalf, is a measure of how unsettled the question of who benefits from AI has become.
For now this is reporting about conversations, not a policy or a deal. The talks are described as preliminary, the share transfers as hypothetical and voluntary, the dividend as one idea among several.
But the fact that the conversation is happening at all marks how far the relationship between the US government and the AI industry has shifted, from arms-length regulation towards something that, if it ever materialised, would make the state a part-owner of the thing it is meant to oversee.
