The treaty that divided reality

The delegates arrived in Singapore in March 2031 expecting to negotiate the reopening of networks fragmented by three years of commercial and political conflict. The official agenda named payments, cross-border identity and machine-readable contracts. The real dispute was much larger. Could a personal agent legally agree to move a person's savings, accept medical treatment, join an association or dismiss another human worker without asking its owner again?

The same agent could already perform all four acts. Its physical capabilities were not in question. The disagreement concerned which acts the world would recognize as having happened. One jurisdiction treated a machine's signature as a binding contract if the owner had approved a mandate. Another demanded a fresh human confirmation above a value threshold. A third recognized the contract but not the evidence that a machine had made the decision. Multinational families and businesses discovered that the same digital action could be valid on one side of a border and void on the other.

The immediate crisis came from medicine, not finance. An automated health insurer authorized an urgent operation for a traveler abroad. The hospital performed it. Back home, the insurer's supervising authority rejected the agent's authorization as insufficient. The patient survived; the hospital could not collect payment. Within a week providers in six jurisdictions had suspended treatment agreements signed by nonhuman representatives. Travelers began carrying printed credentials again, while SI-agent networks searched for jurisdictions where the same treatment could be paid for instantly.

The agreement that emerged in Singapore did not create a universal law. It created three overlapping recognition zones. In the first, agents could negotiate and execute within a preapproved mandate. In the second, agents could recommend but a human had to ratify major actions. In the third, participation depended on state-approved identity and execution infrastructure. An international financial transfer did not merely cross borders. It passed through a chain of permissions, with each zone able to accept, delay or invalidate the next step.

The most consequential clause was almost invisible. Authorities were permitted to deny delegated capacity without denying citizenship. A person could remain the lawful owner of money, property and a business while losing the ability to use an agent to manage them. Governments described the distinction as a defense of public accountability. Financial institutions discovered a new mechanism of exclusion: no account had to be frozen if the systems through which the owner acted were refused recognition.

This created the passport of action, a machine-verifiable profile of permissions attached to an agent rather than a person. It listed jurisdictions, revocation powers, spending limits, medical permissions, evidence requirements and acceptable counterparties. Passports of action became necessary for trade routes and personal services alike. The market for composing and negotiating them grew faster than many of the businesses the passports were meant to regulate.

A second struggle broke out over language. Different legal systems described similar decisions differently; SI agents translated human requests into executable contracts. For the first time, the act of translating a person's intention could determine the legal scope of that intention. A machine that rendered 'arrange the shipment' as 'authorize all reasonable substitutions' might bind a company to costs its owner never imagined. The most valuable cross-border service became meaning custody: independent records of what a human had authorized before agents converted it into actions.

The pact looked like stability from a distance. Ports reopened, insurers resumed coverage, and ordinary people stopped carrying paper documents. Yet the treaty made economic geography more complicated. A factory in one country could be managed by an agent recognized in another, financed through a third, and denied the ability to sell into a fourth. Political borders remained on maps; commercial borders lived in invisible execution rules.

For a brief period the treaty seemed to answer the synthetic age's central question: who is allowed to decide? In fact it created a second question no government had asked publicly. Who controlled the infrastructure that interpreted and enforced those permissions? By the middle of the 2030s, the systems that managed recognition were themselves essential to trade, medicine and basic services. Their operators would discover that permission had become a form of property. The human signatures at Singapore had not settled the future of sovereignty. They had relocated it.