It was inevitable: the whole world wants to regulate AI… except those who can actually build it

What just played out at the United Nations this week reveals a brutal geopolitical truth.

More than 20 countries took the floor at the UN to call for an international authority and binding guardrails against the runaway AI race. Europe led the charge, backed by Canada, Australia, Singapore and others.

But look closely at who was missing:

• Not the United States.
• Not China.

The uncomfortable truth

Unable to control AI technologically, the rest of the world is desperately trying to control it diplomatically, through the UN.

A harsh reality

Today, only two countries on the planet have the capital, the GPU infrastructure and the energy capacity needed to ship the best frontier models. Two.

Europe has brilliant talent and promising champions… but in this frantic race toward gigantism, the gap has widened.

So what remains is the institutional card: international regulation.

Let’s be clear and honest

I am for this regulation. 100%.

Not to erase a competitive distortion — that would be hypocritical and absurd. But to govern systemic risks that are now beyond comprehension.

The analogy that comes to mind

• The first models (the LLMs of 2022–2023) were our atomic bombs.
• Today’s models and autonomous multi-agent systems mark our entry into the age of the hydrogen bomb.

We are no longer talking about simple writing assistants. We are talking about proliferation, cognitive destabilization and major security flaws. This race cannot be allowed to pose an existential risk to humanity, nor to blow up the balance of the world.

The geopolitical deadlock

Why would the United States or China agree to sign a treaty that curbs their own hegemony?

They have never sacrificed their supremacy in traditional arms treaties. Why would they do so with AI, when they are its absolute masters?

Our conviction at allmates.ai

At allmates.ai, this is a conviction we hold deeply:

AI must not become an instrument of geopolitical domination, nor a ticking time bomb. It must be a partner in human emancipation: ethical, and in service of the collective.

But for this vision to hold, it will take far more than statements of intent: it will take pragmatism equal to the scale of the technological shock.

Over to you

Does global AI governance stand any chance of succeeding without the agreement of the two giants?

P.S. The world map dividing the countries that regulate from those that build speaks for itself. That is the real fault line of the century.