Sam Altman, the chief executive of OpenAI, has proposed that the federal government take small stakes in the leading artificial intelligence companies so the public can share in the industry’s financial bonanza. He has reportedly suggested giving the government a 5 percent stake in his own company, which would be worth $42.6 billion at the current market price.

Senator Bernie Sanders of Vermont says the government should take a 50 percent stake in the leading companies, placing the shares in a state-controlled sovereign wealth fund. In addition to a share in the profits, he says ownership would give the government greater control over the development of a technology that has the potential to shake our society.

President Trump also likes the idea of the government holding shares in A.I. companies. “It would be a beautiful thing,” he said recently.

Policymakers ought to resist the temptation. State ownership is not in the public interest.

The government does not need to take financial stakes in companies in order to share in their success. That is the purpose of taxation. As an industry grows, so do the amounts that the government collects in taxes on profits and wages.

Nor does the United States need boardroom seats to participate in decision-making. That is the purpose of regulation. The government has broad powers to check corporate conduct that is antithetical to the public interest.

The A.I. revolution may be so significant as to require a new regulatory framework. It may require a new approach to taxation. But that is where the government should keep its focus.

There are big downsides to government ownership stakes. By providing insulation against market forces, state backing can make companies less competitive, less innovative and less worried about the welfare of their customers. And the history of state investment suggests that, if anything, it often makes regulation more difficult. Governments can be reluctant to enforce rules when action comes at the expense of the state, too.

We also are concerned about the specific dangers that would arise if the current administration took a stake in A.I. companies. Mr. Trump has made clear that he will use almost any tools at his disposal to enrich himself and his allies and to act against Americans whom he perceives as his enemies.

The separation of corporations and state has long been a strength of the U.S. economy. The U.S. government does not own shares in drug companies. It does not own shares in airlines or oil companies or meatpackers. It should not own shares in A.I. companies.

The idea that state ownership is the best way to reap the benefits of corporate success, or to curb its excesses, has an intuitive appeal. It has long been a mainstay of the economic policy of communist nations, and it enjoyed considerable popularity in many industrial democracies during the second half of the 20th century. In the mid-1970s, state-owned companies made up roughly 11 percent of the British economy and employed 8 percent of the nation’s workers.

Such investments are now much less common, however, because the judgment of history was clear. China catalyzed its economic rise by shifting from state ownership to private enterprise. Privately owned American companies outperformed state-owned European rivals in banking, telecommunications, freight transportation and other industries.

In 2023, across the 38 countries in the Organization for Economic Cooperation and Development, companies in which governments held significant stakes accounted for only about 2 percent of the total market capitalization of publicly traded companies.

Our opposition to state ownership is not absolute. Governments and government-owned companies routinely and effectively provide some essential services, including health care and public transportation. Yet these services tend to be those that a market economy fails to provide adequately. A.I. clearly does not fit that description.

The U.S. government also has, at times, taken stakes in troubled companies. For example, during the financial crisis almost two decades ago, Washington bailed out the nation’s major banks, took majority ownership stakes in General Motors and AIG, and took control of Fannie Mae and Freddie Mac. But those kinds of emergency interventions have generally been structured as short-term stabilization measures, undertaken reluctantly and exited as soon as possible.

The proposed investments in A.I. companies are a departure from that history. They would be long-term investments in companies that do not need financial assistance. Indeed, the government would take stakes precisely because the companies are succeeding.

Proposals to give the federal government a stake in A.I. companies often include safeguards to limit the government’s role. Mr. Trump has compared the idea of investing in A.I. companies to the 10 percent stake that his administration took last year in the troubled chipmaker Intel. Those shares are nonvoting, allowing the public to participate in the company’s profits but not in its decision-making.

Even less direct forms of government support, however, can still be problematic. The mortgage finance companies Fannie Mae and Freddie Mac offer an example. The government created the companies, but it did not maintain an ownership stake. Still, in the early 2000s, the companies pursued reckless investment strategies in part because they were confident the government would bail them out, as it eventually did.

The impact on the corporation is only part of the problem. State ownership doesn’t just insulate companies from market forces. It also tends to insulate against regulation.

The public should regard Mr. Altman’s proposal with frank suspicion. It is a payment; the question is what he hopes to receive in return, and the obvious answer is that he is hoping to purchase leniency. He is hoping that the government will deal less strictly with his company if every decision to constrain its ambitions comes, in part, at the expense of the government, too. He has good reason, to judge from the historical record.

State-owned oil companies provide glaring examples. The Norwegian government is a signatory of the Paris Accords, which establish goals for reducing emissions to limit climate change. It is also the majority shareholder in Equinor, Norway’s dominant oil company. Norway’s petroleum wealth provides 30 percent of the government’s annual revenue. Last year, the government blocked an effort by minority shareholders in the company to force Equinor to account for the company’s failure to comply with the environmental policies the government itself has adopted.

Other examples abound. South Africa has legislated more lenient emissions standards for power plants owned by the state utility company, Eskom. Singapore does not require Temasek, a state-owned investment firm, to make the same financial disclosures as many privately owned investment firms. France, for years, exempted some state-owned companies from its bankruptcy laws, allowing those businesses to borrow more cheaply.

Taking stakes in A.I. companies would not increase the government’s ability to control those companies. The opposite is more likely. There is a clear need for legislation to govern A.I., a subject on which we will have more to say in the coming months. Writing those laws will be difficult, but the work is necessary. Ownership is not a substitute for oversight.

The federal government has an important role to play in ensuring that the benefits of A.I. are broadly distributed and in protecting the public from the potential downsides. The best way to achieve these goals, however, is not by picking up the tools of investors, but instead by using the powers of government.