The US government wants equity in AI firms. It also writes the rules for how those AI firms operate. This is not a bug in the regulatory system. It is a feature of a new power structure where the state becomes both investor and overseer. Logic doesn't lie: you cannot audit a system when the auditor owns a piece of the target.
Context
For years, the narrative around AI regulation in the United States followed a familiar pattern: the government would set safety guidelines, enforce antitrust, and ensure fairness—all from a detached, neutral position. The industry would innovate, the state would police. Clean. Simple. Naive.
Now, the Biden administration is quietly floating the idea of the US government taking equity stakes in frontier AI companies. The rationale, as leaked through policy memos, is straightforward: by holding shares, the government gains both a financial return and a seat at the table for strategic oversight. The subtext is even clearer: Washington wants to steer AI development without breaking the private sector's profit motive.
But here is the cold truth: when the government holds equity in an AI firm and simultaneously shapes the regulatory environment for that same sector, a structural conflict of interest is hardcoded into the system. It is the equivalent of a stock exchange letting its own employees trade on insider information—except here the insider is the state, and the information is the law.
Core
Let me reverse-engineer the mechanics. The US government, through agencies like the Defense Department or the International Development Finance Corporation (DFC), would become a shareholder in companies like OpenAI, Anthropic, or even Google DeepMind. As a shareholder, it has a fiduciary duty to maximize the value of its investment. As a regulator, it has a public duty to minimize systemic risk—even if that means imposing costly compliance burdens or forcing a model to be discontinued.
These two duties are mutually exclusive.
Consider a hypothetical: Anthropic discovers a critical safety flaw in its latest model. The government, as a shareholder, faces a decision. It can mandate a recall—hurting Anthropic's stock and its own equity value. Or it can quietly allow a patch to be deployed, keeping the stock stable but exposing the public to residual risk. The incentive to choose the latter is structurally embedded. Volatility is just unpriced risk, but here the risk is not priced—it is hidden behind a political curtain.
From my audit experience during the DeFi Summer of 2020, I saw similar dynamics in DAO treasuries where early investors also governed the protocol. Voter turnout was perpetually below 5%, and whale wallets routinely vetoed security upgrades that would have diluted their holdings. The analogy holds: capital and governance must be separated. Read the code, ignore the roadmap. And here, the roadmap is a regulatory framework written by a shareholder.
But the problem runs deeper than bad incentives. It infects the entire market. Companies that receive government backing will enjoy an unfair competitive moat—preferential access to compute subsidies, relaxed export controls, and a de facto seal of approval that scares off rivals. Companies left out will face a double whammy: higher compliance costs and a skeptical venture capital ecosystem that fears competing against a state-backed giant.
I have seen this pattern before. In 2021, I analyzed 15,000 NFT transactions on OpenSea and found that 85% of wash trading volume was orchestrated by coordinated wallets. The platform had every incentive to look the other way—trading fees were their revenue. When the party that writes the rules benefits from the rule-breaking, enforcement becomes theater. Here, the government is the platform, the AI firms are the trades, and the regulatory framework is the fee schedule.
Contrarian
The bulls might argue that government equity brings stability, long-term capital, and alignment with national security priorities. They have a point. In a world where China's state-backed AI ecosystem receives massive direct funding, the US needs a counterweight. Taking equity rather than nationalizing outright preserves private sector dynamism. Some analysts even suggest that a government shareholder could serve as a patient capital anchor, freeing AI labs from the quarterly profit demands of VCs.
Moreover, proponents claim that conflict of interest can be managed through "Chinese walls"—separate teams for investment and regulation. But in practice, those walls are porous. During the 2022 Terra collapse, I warned a year prior that the dual-token model was mathematically unstable. No regulator listened because no regulator wanted to kill a nascent asset class that donors and politicians had cheered. The same psychology applies here: a regulator who owns equity in an AI firm will find reasons to delay tough enforcement.
Another counterpoint: government equity could force transparency. If the government is a shareholder, it has a right to inspect the code and the training data. That is a net positive for safety. But transparency without independent enforcement is just a PR move. The Securities and Exchange Commission requires public companies to disclose risks, yet we still get fraud. The problem is not the information—it is the aligned incentives to ignore it.
Takeaway
The US government's equity play is a logical next step in the era of sovereign AI, but it introduces a conflict that no smart contract can resolve. If we truly care about safe, responsible AI, we must demand a firewall between the state's role as investor and its role as regulator. Otherwise, we are building a system where the referee owns one team and writes the rules for every other. Logic doesn't lie: that game is rigged from the start. The only open question is whether the market will price in this risk before the first regulatory failure or after.