Singapore proposal for United Nations AI safeguards questions whether server ownership can ensure control in the Gulf

Singapore Foreign Minister Vivian Balakrishnan, in his speech to the United Nations General Assembly on September 26, put forward a proposal to consider drafting a United Nations framework agreement on artificial intelligence safeguards, together with a proposal to establish an institution to support technical standards and verification. Although this proposal remains a topic for international discussion and has not yet become a binding legal commitment, it poses Gulf buyers of intelligent systems with a question that goes beyond the scale of financial spending: how can the institution ensure that the system remains under their full control when multiple international parties are intertwined in its operation?

The ownership of servers does not necessarily mean control over automated decision-making.In practice, an entity in Saudi Arabia, the United Arab Emirates, or neighboring markets can own the physical infrastructure that runs an intelligent service while relying on a foreign model, a separate cloud platform, and a locally integrated system. In such an arrangement, real control cannot be inferred from the servers' location within geographic borders; it is determined by software permissions, contract wording, and strict operating procedures. Singapore’s statement emphasized human accountability, reliable evaluation, and the ability to intervene immediately when autonomous systems take unexpected paths.

Regionally, the UAE charter for developing and using artificial intelligence, which defines safety, privacy, and governance among its core principles, offers a practical point of comparison; national charters and the proposed international frameworks call for scrutiny of how public commitments are translated into live systems operating on the ground. The particularity of publishing in Arabic also emerges here: a system that achieves acceptable results in English evaluations needs an independent, intensive review before it engages with Arabic-language users, because local dialects, multilingual texts, precise terminology, and the consequences of linguistic errors affect test design.

Reality demands a distinction between drafting precision and implementation suitability. An intelligent public-service assistant that explains application procedures to users while also having the authority to submit forms and modify records poses a challenge that goes beyond language alone; the buyer needs to know when the system should request human confirmation, when it should refer a case to a specialist staff member, and how any erroneous submission can be traced and corrected. The existence of a general safety document from a foreign supplier does not guarantee coverage of the buyer’s specific tasks, and while shared international evaluation methods have facilitated supplier comparison, they do not ensure testing of every local use case nor resolve issues of data access or contractual liability.

What is changing today for technology procurement managers and engineering teams in the Gulf, Egypt, and the region is not tied to waiting for a United Nations treaty that may take years, but to imposing stricter audit conditions when contracting. The direct step requires identifying gaps between the model developer, the infrastructure operator, and the final application developer. This includes documenting the party authorized to halt automated processes across the entire service flow, requiring evaluation evidence that covers tasks in Arabic and their operational context, and pre-agreeing on a joint investigation mechanism when technical incidents arise with responsibility shared among multiple suppliers. These discussions also place investors before a test of their emerging companies: does the company rely on safety assurances from a foreign supplier, and does it have the capacity to conduct independent investigations of failures if its relationship with that partner ends?

In the diplomatic and regulatory arena, the United Nations’ global dialogue on AI governance provides an established forum for debate, but the real impact of regional policies will not be achieved merely by hosting conferences or signing joint statements; it requires direct participation in technical working groups and the provision of evidence from the operation of Arabic services and complex procurement contracts. The commercial benefit sought for the region lies in reducing uncertainty between buyers and suppliers, clarifying who holds decision-making authority, who retains audit evidence, and who bears responsibility for immediate intervention when an error occurs.