Last month, U.S. President Donald Trump suggested the government buy equity in AI companies, in response to concerns that the industry’s boom is leaving most Americans behind. His political opposite, Senator Bernie Sanders, had earlier proposed creating a sovereign wealth fund with up to a 50% stake in AI companies, and OpenAI is reported to be in talks to give the U.S. government a 5% stake as it prepares for an IPO. But is a government stake in AI companies necessary? And is it wise?

A government owning equity in a company is not a new concept. In many countries, the government owns stakes in companies seen as critical to the economy or to national security. These may include oil and gas companies, iron and steel manufacturers, banks, telecom firms, aircraft makers, ship builders, and others.

The U.S. has a precedent in the Alaska Permanent Fund, which pays dividends to citizens from the state’s oil wealth. In the past year, the Trump administration has made deals to buy equity stakes in more than two dozen firms across industries, including semiconductors, nuclear energy, minerals, quantum computers, and steel. OpenAI and Anthropic have both proposed a public or sovereign wealth fund to distribute wealth created by AI among the public.

But the developers of ChatGPT and Claude are very different from steel manufacturers, and owning a minority stake in a company does not amount to public ownership, and does not mean that the American public will benefit from the company’s profits. AI is expected to transform every industry, touch every aspect of our lives. The U.S. has no federal AI laws, which the current administration believes would stifle innovation and blunt its edge against China. If the government held stakes in AI companies, there may be even less incentive to intervene on safety, antitrust, and content regulation, or take any regulatory action that may affect their market value.

There may also be other conflicts of interest. Should users be worried about privacy and government surveillance? Will data centers be greenlit without public consultations or state approval? What about the growing pile of lawsuits against AI companies — would users and states still be able to sue AI companies if the government held stakes in them? When Apple and OpenAI go to court, whose side will the government take? And what if their revenues fall — will government ownership make these firms “too big to fail?” It’s a very slippery slope.

When the government becomes a shareholder in a private-sector entity, “politics trump profits, favoritism and cronyism take root, innovation suffers, competitiveness erodes, and regulation is corrupted,” former presidential candidate Michael Bloomberg wrote this week, calling it “a dangerous idea.”

It’s a divisive issue. Mona Sloane and Emanuel Moss, professors of data science at the University of Virginia, argued in a recent paper that since AI systems “are infrastructures that intersect with the public interest in vitally important ways,” they should be reframed as a public utility. This will ensure public accountability and establish democratically governed AI infrastructure, they said. 

Curiously, in exploring ownership in AI companies, the U.S. is considering a path already taken by China. Beijing holds stakes in major tech companies, known as “golden shares,” which give it special voting power and veto rights over corporate decisions. China’s state-backed AI industry fund plans to invest in one of its best-known frontier companies, DeepSeek.

But the Chinese government’s ownership in AI companies is largely about self-sufficiency. The national AI fund targets firms across the entire ecosystem, from chip design to AI applications, and its investments are aimed at having companies align with the country’s strategic interests. China also offers incentives and subsidies while regulating strictly — and quickly — on matters such as safety, content, and export controls to ensure AI that is “secure and controllable,” as President Xi Jinping said last week.

Americans are not as excited about AI as others across the world. The unease is likely set to grow as upcoming IPOs make a small handful in the AI industry extremely wealthy. Trump is right to worry that the benefits are not being shared equitably with the broader public. But a government stake in AI companies is not the answer. Instead, a government fund like China’s that invests in startups, or an AI safety institute like Singapore’s or the U.K.’s, is a better bet for the industry — and the public.

What I'm reading

• A profile of Stanford University economist Erik Brynjolfsson, who is tracking AI’s effects on workers, firms, and human welfare.
   
• A paper documenting the use of humanoid robots to surgically remove gallbladders from living animals, with human surgeons remotely controlling the robots’ movements, in an unprecedented medical experiment.
   
• The Humanist Review of AI, a new quarterly publication funded by Microsoft AI, which features experts from different fields on the impact of AI on society. The first issue includes essays by philosopher Alain de Botton, economist Daron Acemoglu, and Taiwan’s cyber ambassador Audrey Tang.

And one more thing...

How did you mark the 250th birthday of the founding of the U.S.? Podcaster Roman Mars — who hosts one of my favorite podcasts, 99% Invisible — is doing it with a series called A History of the United States in 100 Objects. Produced by BBC Studios, which created the podcast A History of the World in 100 Objects, the new series has Mars exploring objects that are less widely known.

There’s a gold coin from an 1857 shipwreck that triggered a financial panic, a ceramic Dalmatian from the set of Wheel of Fortune, and a tiny screw that shows how the U.S. created a hidden industrial empire by gradually imposing a single standard in production lines across the world. Do you have an object that you think merits a mention? Let us know.