The AI Frontier Non-Proliferation Treaty: Hegemony, State Sovereignty, and the New Global Access Order

The development of the most capable artificial intelligence systems is currently concentrated in the United States and China. Both possess complete frontier complexes, and their relative positions in terms of capability oscillate. In relation to the rest of the world, however, a structural asymmetry exists. This asymmetry does not arise solely from more capable models. The frontier itself is an ongoing process of development in which new insights feed directly into the next generation of systems. Actors outside this process must therefore attempt to catch up with a leading edge that is itself continuously advancing. From this structure, the article develops the thesis of a possible AI Frontier Non-Proliferation Treaty. The United States and China treat the AI frontier as a central strategic asset. Despite their rivalry, both therefore have an interest in limiting the emergence of additional autonomous frontier complexes. A joint delimitation of the protected leading edge could prevent competition for international partners from compelling both sides to offer increasingly extensive forms of access. The Treaty on the Non-Proliferation of Nuclear Weapons serves as the historical reference model for the durable institutionalisation of an existing technological asymmetry. The resulting power extends beyond control over a strategic technology. The frontier hegemons control the most capable systems through which globally available knowledge is evaluated, condensed, and recombined. Their outputs shape human and institutional action and alter the reality from which new knowledge subsequently emerges. An AI Frontier Non-Proliferation Treaty would therefore do more than organise access to the highest available intelligence. It would consolidate a structure of power that increasingly feeds back into the further development of the global knowledge space.