AI Should Answer to Users, Not Washington

Vice President JD Vance recently offered a stark warning about artificial intelligence. AI will “unleash a lot of wealth creation,” he told Joe Rogan on his show recently, but “if that wealth creation all goes to some segment of people, you’re going to have communism.”

If the concern is that AI-generated wealth will accrue to too few people, one response is to allow for expanded private ownership. Another is to have the government own the companies instead. It is the latter that has long been the central economic prescription of communism.

And that history makes current proposals from the administration for federal ownership stakes in the AI industry especially hard to square with his concern.

President Trump is exploring ways to give the public a financial stake in leading AI companies. Sen. Bernie Sanders (I‑Vt.), meanwhile, has gone much further, calling for the government to own half of the largest AI firms and to receive board representation.

But if Washington takes an ownership stake in the companies behind these tools, it risks changing what AI companies actually compete for. Instead of answering first to users, they will have a powerful new incentive to answer to politics.

Government ownership would not merely distribute gains. It would reshape the incentives that determine which companies succeed and shape the future of technology to bureaucrats’ whims.

Indeed, claiming a piece of the AI industry outright would mark a dramatic escalation in Washington’s burgeoning push to take ownership stakes in private companies. The federal government already shapes AI through standards and guidance, procurement rules, and export controls. Make the federal government a shareholder and it gains a direct financial interest in the market value and market share of particular firms.

In effect, policymakers are looking to create a government that is simultaneously regulator, shareholder and policymaker in the AI industry — hardly a neutral referee. 

Once Washington owns part of a company, neutral decisions become harder to separate from the government’s ownership interest. Competitors may wonder whether regulatory decisions and procurement choices are based on objective standards or maximizing the government’s investment.

During the auto bailout of 2008 and 2009, the Treasury Department became a controlling shareholder in GM and a direct shareholder in Chrysler after their bankruptcy restructurings, exposing both companies to political lobbying over dealership and facility closures. But those stakes were crisis measures to be unwound as soon as practicable; today’s AI ownership proposals would make Washington an owner of cutting-edge companies as a matter of policy.

Even if officials insist they would act objectively, the market would still hear a different message. The problem is not simply corruption or bad intentions — it’s the incentives. Companies would learn that political favor can matter as much as technical merit. Investors would look not only for better products, stronger teams, or more efficient models, but for signs that a firm has Washington’s blessing. Startups and new entrants would face pressure to hire lobbyists, cultivate the right officials, and structure themselves around political access.

Unlike the American auto industry, AI is still young and dynamic. Trying to identify today’s permanent winners is like trying to pick the internet giants of 2026 from the vantage point of 1996. Government ownership would freeze favored players into a preferential market position at a time when there should be robust competition.

The risks in AI extend beyond the economics. AI is a powerful technology for accessing information. Government investment could impact how companies choose to provide users with information or how they develop their models lest they lose favored status.

Recent experience shows how disruptive political decisions can already be for AI users. In June, after the Trump administration abruptly imposed export controls on Anthropic’s Claude Fable 5 and Mythos 5, Anthropic suspended access to both models for all users. The episode illustrated the broader risk that a sudden government action can change the availability and reliability of tools businesses increasingly depend on. Government ownership would add another layer of political uncertainty.

There is a better path. Congress should reject federal ownership stakes in AI companies and clarify the limits of existing authorities so that executive agencies cannot use regulation, procurement, export controls, or informal pressure to pick winners. Agencies and industry can develop transparent, voluntary frameworks that address legitimate safety and security concerns without turning Washington into a market participant.

The heart of the issue is simple: Will AI winners be chosen by users or by Washington?

The answer should be obvious. The greatest risk to AI isn’t so much that government fails to pick the right winner but that it gets the power to pick them at all.