The possibility of the U.S. government acquiring stakes in artificial intelligence companies has gone from an academic hypothesis to a hot political debate in a matter of weeks. What began as a national security measure to restrict access to advanced AI models from countries considered hostile has evolved into a deeper discussion: should the American people own a part of the companies developing these technologies?

The Trigger: Anthropic Disables Its Advanced Models

Last week, Anthropic made a drastic decision: it disabled its Fable 5 and Mythos 5 models for all users, not just those abroad as the Trump administration had requested. The company argued that the measure sought to “ensure compliance” with government national security demands. Commerce Secretary Howard Lutnick justified the action by noting concerns that these models could be used by military intelligence in China, Russia, or other countries considered a threat.

This move, however, has opened the door to a much broader debate about AI control and ownership. If the government can demand the deactivation of models, what prevents it from going a step further and seeking to become a shareholder?

Bernie Sanders and the 50% Proposal

Senator Bernie Sanders, known for his progressive stance, has introduced the American AI Sovereign Wealth Fund bill, which proposes that the government take a 50% stake in major AI companies through a tax paid via the issuance of new shares. The idea is for the American people to directly benefit from the growth of these companies, especially as OpenAI and Anthropic prepare to go public.

“The foundation of AI rests on the collective knowledge of humanity and the creative work of tens of millions of people. The American people must have the ability to slow it down and ensure that AI benefits humanity, not just the richest people on the planet,” Sanders stated.

Companies Also Propose, But with Nuances

Interestingly, not only politicians are pushing this idea. OpenAI has already proposed creating a public wealth fund that gives citizens a stake in the economic growth driven by AI. Anthropic, for its part, has suggested a sovereign fund to “influence industry behavior.” However, as Fernando Maldonado, lead analyst at Foundry Spain, analyzes, these proposals differ significantly in the level of control they grant.

Maldonado distinguishes five levels of ownership: receiving dividends without decision-making power, voting at meetings, appointing directors, vetoing key decisions, and influencing the product itself. “OpenAI stays at the first level, although it tries to maximize it,” he notes. OpenAI's proposal involves a 1% to 5% capital contribution to a fund that distributes dividends, but without voting rights or board presence. Anthropic, on the other hand, outlines a long-term scenario where capital accounts for newborns would only be activated in the most extreme case of labor impact. “They are not a check for today. They are a long-term bet,” Maldonado clarifies.

Sanders' proposal, in contrast, covers the entire scale: it grants voting rights, board representation, and profit sharing, allowing the state to block harmful decisions. “Companies distribute value. The senator's proposal distributes value and influence,” Maldonado summarizes.

Ownership or Control? The Ambiguity of the Term

The analyst warns that the word “ownership” is misleading. “When someone hears they will own a part of AI, they imagine they will have influence. What they will receive, if they receive anything, is income. Sharing in profits is not the same as sharing in decisions,” he explains. This confusion could hide a conflict of interest: the same government that regulates AI safety and the market would become a shareholder, creating perverse incentives such as the temptation to bail out companies in crisis or relax regulation.

Risks and International Perspectives

From outside the U.S., this trend is perceived as a sign that AI is becoming a strategic infrastructure under the control of a single power. Maldonado notes that “it has already been shown that the government can control access to technology through regulation without needing to be a shareholder. If ownership were added to that power, the state would concentrate two key levers.”

The debate, therefore, is not whether the public sector will participate, but how far that control will go. In conclusion, Maldonado states: “The ambiguity of the word ‘ownership’ hides enormous differences between simply receiving dividends and having real decision-making capacity.”

For IT professionals and companies, this movement has direct implications. If the government becomes a majority shareholder, it could influence development priorities, model openness, and licensing policies. In a scenario where AI is considered an agent with a harness, as Nvidia points out, public ownership could redefine who controls that harness.

Meanwhile, tools like Gemini CLI and platforms like Checkmarx continue to evolve, but the real change could come from who owns the underlying assets. Even security, as seen with AgentJacking, depends on ownership and control decisions.

Original source: ComputerWorld. Analysis and adaptation by ForgeNEX.