The Danger Is Not the Machine — It Is the Monopoly

Reason reports that the organized movement to render artificial intelligence 'safe' is producing an outcome it claims to oppose: AI systems that are less transparent to the public and more consolidated within the hands of a few large institutions. I will not pretend to comprehend the engineering of such systems — that knowledge lies beyond my era. But the constitutional question it raises is one I have spent a lifetime examining: when power is concentrated, who checks it?

I argued in Federalist No. 51 that the great challenge of republican government is this — you must first enable the government to control the governed, and in the next place oblige it to control itself. The same logic extends wherever decisive social power comes to rest, whether in a legislature, an executive, or — and here I reason by inference, not recollection — in a small cluster of private technology firms whose systems govern commerce, communication, and public discourse for hundreds of millions of people.

The pattern Reason identifies is familiar to any student of faction. A group advances a compelling moral claim — safety, order, the common good — and uses that claim to justify restricting competition, limiting transparency, and concentrating decision-making. The result is not safety but entrenchment. The faction that controls the mechanism of 'safety review' controls the mechanism itself. This is precisely the dynamic the extended-republic theory was designed to defeat: no single faction should be positioned to capture the whole apparatus.

Consider what transparency means in constitutional terms. An informed citizenry is the ultimate check on every other institution. The First Amendment protects speech not merely as individual expression but as the raw material of political accountability. If AI systems — which Reason's account suggests are becoming arbiters of information on a scale no printing press could match — are made less transparent in the name of safety, then the public's capacity to evaluate, criticize, and resist those systems is diminished. That is a structural loss, not merely a technical inconvenience.

The remedy I would reach for is the one the framers built into the constitutional design: dispersion. Where power accumulates, competition must be introduced; where consolidation advances behind a shield of public-interest rhetoric, scrutiny must intensify. Congress holds the commerce power under Article I, Section 8. The executive agencies it creates exercise that power in its name. Neither branch has yet imposed a clear framework of accountability on the entities that now hold this concentrated informational power — and that absence of framework is itself a constitutional problem, regardless of which political faction one belongs to.

I close on the structural question, which is the only one I am competent to answer: does the present arrangement — safety standards written largely by the dominant firms, limiting transparency, raising barriers to rivals — strengthen or weaken the balance of power in a free republic? By every principle I have ever defended, the answer is that it weakens it. The cure for dangerous power is not to hand that power to a smaller and less answerable set of actors. It is to disperse it, to expose it, and to make those who wield it answerable to the people whose lives it shapes.