A global AI safety regime should learn from nuclear arms control by focusing on the physical infrastructure that enables strategic capabilities. Instead of just seeking promises, it should aim to control access to chokepoints like advanced chip manufacturing and the massive data centers required for frontier models.

The belief that AI development is unstoppable ignores history. Global treaties successfully limited nuclear proliferation, phased out ozone-depleting CFCs, and banned blinding lasers. These precedents prove that coordinated international action can steer powerful technologies away from the worst outcomes.

The popular comparison of AI to nuclear weapons has a critical flaw. Nuclear regulation relies on tracking scarce, physical, and interceptable fissionable materials. AI, as software and weights, can be copied and distributed far more easily, making the nuclear non-proliferation playbook a poor and dangerous model for AI governance.

The US nuclear weapons industry operates as a hybrid: the government owns the IP and facilities, but private contractors like Honeywell and Boeing operate them and build delivery systems. This established public-private partnership model could be applied to manage the risks of powerful, privately-developed AI.

Instead of broad, paralyzing regulation that could stifle innovation, a more effective approach is to create a clear distinction between AI with weaponization potential and AI for civilian use. This mirrors the dual-use framework for nuclear technology, allowing progress while managing existential risks.

Ben Horowitz revealed that Biden administration officials defended the idea of regulating AI—which he framed as "regulating math"—by citing the precedent of classifying nuclear physics in the 1940s. This suggests a governmental willingness to treat core algorithms as controlled, classifiable technology, potentially stifling open innovation.

Comparing AI to a nuclear weapon is misleading because AI is a general-purpose technology, not a single-use weapon. A better analogy is the Industrial Revolution. Society didn't give governments control over industrialization; it regulated specific dangerous end-uses like chemical weapons. Similarly, we should ban specific destructive AI applications, not the underlying technology.

The history of nuclear power, where regulation transformed an exponential growth curve into a flat S-curve, serves as a powerful warning for AI. This suggests that AI's biggest long-term hurdle may not be technical limits but regulatory intervention that stifles its potential for a "fast takeoff," effectively regulating it out of rapid adoption.

International AI treaties are feasible. Just as nuclear arms control monitors uranium and plutonium, AI governance can monitor the choke point for advanced AI: high-end compute chips from companies like NVIDIA. Tracking the global distribution of these chips could verify compliance with development limits.