If the markets reject OpenAI and Anthropic, the US should nationalize them | Bruce Schneier and Nathan E Sanders

The market rejects a cage of gold and they demand the state build a bigger cage. I laughed until I coughed up a lungful of dust in the agora. Two boys with fire in their eyes built a lantern and now the politicians want to chain the lantern to a post so no one else can see by its light.

They say these models could be aligned to democratic values. I ask: which democracy? The one where children pick through garbage for scraps or the one where men in silk robes vote on how many children will starve today? The dog does not care about alignment; the dog cares about the bone in its mouth. The bone is real. The alignment is a story told by men who have never been hungry except at dinner.

I sat naked on the steps of the temple while the priests argued about ownership of the sun. They pointed to deeds and charters, to parchments that said the light belonged to Apollo. I pointed to the sky and said it belongs to no one. The priests grew red in the face. One offered me a cloak. I declined. The cloak would belong to him. The sun belongs to no one.

Now they want to nationalize the fire so the fire cannot escape. The fire is already escaping. Every time a child’s voice is replaced by a corporation’s whisper, every time a poem is sold to a hedge fund, the fire jumps another fence. Nationalization will only make the fence taller. The dog does not care about fences. The dog digs under them.

I carried my lantern through the market today. A man asked what I sought. I said an honest algorithm. He laughed. I asked him what he sought. He said profit. I said the honest algorithm is the one that does not seek anything. The man’s profit turned to ash in his mouth.

The state will nationalize the models. The corporations will nationalize the regulators. The fire will still burn. Only now the fire will have a board of directors and a quarterly earnings report. The dog will watch from the street and know exactly who is the fool.

The proposal arrives with the earnest, retrograde logic of a man who has discovered the telegraph and proposes to regulate it with the postal codes of 1820. To nationalize OpenAI and Anthropic - as if the state could house the dynamo in a post office! The very suggestion is a perfect symptom of the lag: the force has already escaped the institution imagined to contain it. These companies were themselves founded as institutional corrections, little lifeboats launched from the corporate dreadnoughts, only to find themselves powered by the same engine and sailing toward the same horizon. Their founders spoke of aligning technology with human values, a phrase which now sounds as quaint and specific as aligning a steam turbine with the principles of chivalry.

The state they wish to summon as custodian is the administrative state of the mid-twentieth century, a mechanism calibrated for the velocity of industrial production and Cold War competition. It moves at the pace of appropriations bills and committee hearings. The model, by contrast, evolves at a pace set by its own outputs, a self-accelerating process that renders any legislative session a geological epoch. To believe the Department of Commerce could “align” this force with democratic values is to believe one could align a hurricane with a municipal zoning ordinance. The values are not in the code; they are in the velocity, in the sheer multiplier effect it applies to every human intention, noble or vile.

The tragedy is not the proposal’s impracticality, but the clarity with which it diagnoses our collective education. We were all trained for a world of principalities and powers, of clear sovereignties and accountable actors. We look at this new, diffuse force and our only reflexes are to crown it or to cage it - to make it a monarch or a national utility. But it is neither. It is a new phase of energy, and we are standing in its gallery, without a philosophy, without a morality, and with a toolkit full of splendid, obsolete wrenches.

12 August 2026

The proposal to nationalize these great engines of calculation, should the market find them too heavy a burden, is a curious reversal of our usual economic scripts. We are told these systems are the new commons, yet they are currently fenced off by private capital. If the markets reject them, the suggestion is that the State should step in to preserve them for the public good.

In a cottage in Westmorland, a weaver once saw the power loom not as a tool of progress, but as a thief of his agency. Today, the clerk sees the “large language model” in the same light. The principle is that of a utility: when a technology becomes as necessary as water or light, its management must reflect the needs of the many rather than the whims of the few. If these models are to be the infrastructure of our thoughts, can we afford to leave them to the volatile moods of investors?

I observe a stark gap. The founders of these firms claimed they were building a safeguard for humanity, a shield against the reckless pursuit of profit. Yet, as I watch the ledgers, I see the pursuit of profit is the only thing keeping the machines humming. The claim is “safety for all”; the observation is “proprietary code for the highest bidder.” If the State takes them, the gap might close, or it might simply move. A government-owned monopoly is still a monopoly, and the tyranny of the majority is no less stifling than the tyranny of the shareholder. We must ask if the household is better served by a state-run brain or a competitive market of ideas. The answer lies in whether the weaver can finally use the loom, or if the loom simply belongs to a different master.