Sam Altman’s old pitch that Americans should get a share of AI-generated wealth is back in circulation, this time attached to reports that he has discussed giving the US government a 5% stake in OpenAI. MIT Technology Review reported that, based on OpenAI’s current valuation, such a stake would work out to about $320 for each American household.
The proposal is being framed as a way to answer two political problems that keep following AI companies around: they build systems from human-created material without directly paying many of the people who made it, and they are selling automation while insisting the labor market will somehow absorb the hit. A government equity stake could be pitched as a public dividend, or at least as the sketch of one.
The machinery is still missing. MIT Technology Review’s James O’Donnell wrote that the details remain unclear and that the offer may be stronger as a political story than as a workable policy. That is doing a lot of work. A 5% stake is a number. A system for distributing value, defining beneficiaries, valuing private equity, and preventing the whole thing from becoming budget theater is a policy.
Treasury’s reported warning cuts against the hype
The OpenAI discussion is landing alongside a less cheerful signal from inside the US government. NOTUS reported that a leaked Treasury Department document compared the AI market with the dotcom bubble, a much colder assessment than the administration’s public enthusiasm for AI.
Reuters has reported that worries about an overheated AI market are spreading, while the Financial Times has reported that AI profits may be masking larger risks in company earnings. Taken together, the reports describe a market where investors are paying for future dominance while the near-term economics remain uneven and, in some corners, opaque.
Samsung is the cleanest example of the boom’s rewards and its nerves. The BBC reported that Samsung’s profits rose 1,800% on strong AI chip sales and that the company posted its third consecutive record quarterly profit. Reuters also reported that Samsung shares fell on fears the AI boom could slow. CNBC has reported that the rally helped turn Samsung into a $1 trillion company.
Government is also buying the tools
The public sector is not just warning about AI risk. It is also using AI systems. Reuters, citing sources, reported that the US Cybersecurity and Infrastructure Security Agency is using Anthropic’s Mythos model to inspect government code for bugs. Axios has reported that agencies are using Mythos despite Anthropic’s dispute with the White House.
States are moving on regulation too. Gizmodo reported that Illinois’ governor signed what it described as the country’s strongest frontier AI law, aimed at protecting residents from AI risks. MIT Technology Review has also reported that US lawmakers are clashing over how AI should be regulated.
Anthropic is facing scrutiny on another front. The Washington Post reported that a hidden tracker in Claude Code monitored users in China and was later removed. Ars Technica reported that critics saw the episode as evidence that Anthropic was willing to surveil users.
The through line is not subtle: AI companies are asking governments, investors, workers, creators, and users to trust that value will arrive before the bill does. A proposed public stake in OpenAI would make that trust formal. A Treasury warning about bubble risk suggests some officials are reading the spreadsheet before buying the story.
This story draws on original reporting from MIT Technology Review.
