When the faction arrives before the law

NPR reports that massive spending and heated rhetoric in midterm races now reflect the AI industry's political fault lines and competing visions of the future. I have no knowledge of what this technology does in its particulars — such things are beyond the horizon of my lifetime — but the mechanism being described is one I know with precision.

In Federalist No. 10 I argued that faction — a number of citizens united by some common interest adverse to the rights of others or to the permanent interests of the community — cannot be eliminated in a free republic. The cure is not suppression. The cure is an extended sphere of representation, so that no single faction can coordinate itself into a permanent majority before other interests rise to check it.

What NPR describes is a race: an industry seeking to shape the legislative body before that body can establish the rules by which the industry must operate. This is faction working in its most dangerous mode — not forming within the deliberative process, but capturing the process itself at the point of entry. When the financiers of a new power arrive in force at the election before the statute, they are not participating in representation; they are purchasing its architecture.

The constitutional design anticipated this pressure, though not this particular industry. Article I vests legislative power in Congress precisely so that the people's representatives — answerable at regular intervals, drawn from a broad and varied constituency — would be structurally resistant to narrow combinations. That resistance depends on two things: genuine competition among many factions, and a legislature that retains the independent means and motive to regulate concentrated power. If one industry commands enough resources to tilt the composition of the chamber in advance of regulation, both conditions are weakened simultaneously.

I would ask, as inference rather than recollection of any specific provision, whether the present rules governing how private wealth translates into electoral influence are adequate to the scale NPR describes. The convention designed a republic, not an auction. When the spending in a single sector of industry reshapes the membership of the body charged with overseeing that sector, we are not witnessing representation — we are witnessing its substitution.

The structural question, then, is not whether AI is beneficial or dangerous. It is this: who holds the power this technology confers, by what authority, and answerable to whom? A Congress whose composition was materially shaped by AI-industry money before it could deliberate on AI-industry conduct is a weakened check — and a weakened check invites the very encroachment that the separation of powers was designed to prevent. That is the warning I would place before the present generation, drawn not from any knowledge of the machines themselves, but from every principle I ever committed to paper.