The United Nations has discovered artificial intelligence. The machines, unfortunately, discovered the United Nations some time ago. That is the uncomfortable joke behind the grand discussions now surrounding AI governance. World leaders gather in New York, pronounce the words security, responsibility, humanity, and international cooperation, and then return to governments that still struggle to regulate yesterday’s technology. Meanwhile, artificial intelligence is moving at a speed that makes diplomatic procedure look almost geological.
The problem is not that the UN is discussing AI, it absolutely must. The problem is that much of the conversation risks being conducted through the mental architecture of the twentieth century: identify the dangerous technology, establish rules, create monitoring mechanisms, and hope sovereign states behave themselves. It worked, imperfectly, for nuclear weapons. AI is something else entirely.
A nuclear weapon requires extraordinary physical infrastructure. AI requires computing power, energy, data, talent, chips, and access to networks. These resources are not distributed remotely equally.
The UN itself has acknowledged this enormous capacity gap. Its current discussions include proposals for stronger access to infrastructure, skills, data, and technology, whilst a recent UN report highlighted fragmented financing for AI capacity-building in developing countries. The first Global Dialogue on AI Governance was deliberately designed to bring the Global South into the conversation.
That is welcome. But participation is not the same thing as power. A country can have a seat at the diplomatic table and still have no semiconductor industry, no hyperscale data centres, inadequate electricity grids, limited research capacity, and little influence over the companies developing the most powerful models.
This is where the AI debate becomes uncomfortable. The great technological powers are not merely developing products; they are building the infrastructure upon which everyone else may eventually depend. The danger is therefore not only that AI might become autonomous, malicious, or uncontrollable. The quieter danger is dependency.
The Global South could become the customer, the data source, the testing ground, and eventually the labour market of an AI economy designed somewhere else. That would be an old colonial pattern wearing a new digital suit.
This is why the UN cannot allow AI governance to become another exclusive conversation among governments, technology corporations, and the countries that possess the money to build the machines. António Guterres has warned that power is increasingly shifting towards a handful of private corporations and individuals, whilst global institutions still reflect an older distribution of power. That observation goes directly to the heart of the AI problem.
There is also a delicious contradiction here. Some governments argue that global regulation would interfere with innovation. Others demand stronger international rules. Technology companies themselves increasingly acknowledge that no single country or company can manage the technology alone. At the same time, the companies possessing the greatest technical capacity remain overwhelmingly concentrated in a small number of states.
So who exactly is writing the rules? And for whom?
AI governance cannot simply become nuclear non-proliferation with better software. The objective should not merely be preventing catastrophic misuse by powerful states; it must also prevent the creation of a permanent technological hierarchy in which a handful of countries own the intelligence and everybody else rents it.
The UN therefore needs to talk about infrastructure with the same seriousness with which it talks about ethics. Electricity, broadband, data centres, computing capacity, education, open research, local-language systems, and technical skills are not peripheral development issues, they are the foundations of technological sovereignty.
Otherwise, we may succeed spectacularly in regulating the future whilst leaving most of humanity without any meaningful ownership of it. That would not be global governance. It would simply be global dependency with a United Nations logo on the paperwork. The machines will not care who writes the rules. People should.
* * * * * * * * * *
Published in Ovi magazine and across diverse print & digital platforms.
