Sanders Urges AI Giants To Halt Development
The senator’s request arrived on a timeline that assumed institutions capable of processing it at the speed they last achieved when the Interstate Commerce Commission was learning to regulate railroads - a body deliberating in sessions, hearings, and printed reports, against a subject that revises itself in the interval between hearings. Bernie Sanders has asked Meta, OpenAI, and Anthropic to halt their development of artificial intelligence. The request is not unreasonable. It is simply calibrated to an instrument that no longer exists - the pause, the moratorium, the gentleman’s agreement to wait while wiser heads convene. Wiser heads convening was a nineteenth-century technology. It required all parties to move at comparable speed, and comparable speed was the dynamo’s first casualty.
Consider what the senator is actually asking three organizations to do: cease being what they are. Meta, OpenAI, and Anthropic did not become what they are by pausing; they became what they are by the specific mechanism of not pausing, of treating each competitor’s advance as proof that pausing is a form of self-liquidation. This is not cynicism on their part - it is arithmetic. If Anthropic halts and OpenAI does not, Anthropic has not made the world safer; it has made itself irrelevant to the world’s safety, ceding the decision entirely to whoever kept moving. The senator is asking each dynamo to stop generating power on the assurance that the other dynamos will do likewise, which is precisely the assurance that no dynamo in history has ever been able to give, because the dynamo’s entire value proposition is that it does not stop.
The entropy here is worth measuring precisely, because it is not the entropy of malice but of asymmetry. The Senate operates on a cadence of hearings, markup sessions, cloture votes - a cadence designed, reasonably, for an eighteenth-century republic’s most urgent business, which was slow by design, since slowness was thought a virtue against faction. The laboratories operate on a cadence of model releases measured in months, sometimes weeks, each release incorporating capability the previous cadence had not anticipated regulating. The distance between these two clocks is the actual subject of this event, more than the specific technology under discussion. Sanders is not wrong that a risk exists. He is measuring the risk with an instrument - the halt, the moratorium - built for an era when technological advance moved slowly enough that a request to stop was a request the target could actually fulfill without being immediately supplanted.
I think here of a scene: a shop foreman in 1880, watching a new loom arrive that runs three times the speed of the one it replaces, and remarking to no one in particular that the factory’s safety inspector still schedules his rounds by the old machine’s rhythm. The inspector is not incompetent. He is uncalibrated. The three laboratories now named by the senator are not villains in this scene; they are the new looms, running at the speed the market has told them is survival, while the inspector’s schedule remains fixed to an earlier machine’s pace.
What the mismatch measures, finally, is not whether artificial intelligence is dangerous - that question exceeds what any single dispatch can settle - but whether the halt is a mechanism that still functions at all when addressed to competitive actors rather than to a single sovereign body. A king can be asked to halt his armies, because he alone commands them. A senator asking three competing companies to halt simultaneously is asking for a coordination that only regulation, not appeal, has ever reliably produced - and regulation, in this instance, has not yet been written, only threatened. The threat is real. The writing is not. Between the threat and the writing lies the interval in which the dynamo, unregulated, does what dynamos do: it turns.