In the first quarter of 2026, venture capital poured $242 billion into artificial intelligence companies, roughly 80 percent of all the venture money in the world that quarter, according to Crunchbase. Beside that number, the money we spend making these systems safe almost disappears. Open Philanthropy, the largest philanthropic funder in the field, gave around $50 million to technical AI safety in 2024. By one widely cited estimate, the whole safety sector, venture capital included, raised about $3.9 billion in the four years to 2025. We spend a fortune in a single quarter to build the thing, and we spend loose change, over years, to ask whether we are building it well or should be building it at all.

A government's documents declare its culture, while its budgets declare its reality. The distance between the two is a confession, and anyone who works inside the state knows that this gap is where the real story lives.

While the world argues about the ethics of these machines it argues in languages that never suspect how different they are. One capital speaks of rights and another of innovation, one of harmony and another of existential risk, and each believes it is part of a single global conversation when it is not. What we politely call the AI ethics debate is really a set of regional dialects, each carrying a vocabulary of its own and a blind spot it has learned never to name. It is worth listening to them one by one, to hear what each of them cannot bring itself to admit.

Brussels speaks the language of rights, and its AI Act is the first genuinely comprehensive body of law built around risk and fundamental rights. What it cannot see is that it regulates a market in which it does not compete, because the frontier models that set the terms are trained elsewhere and reach Europe already shaped by choices made in another tongue.

Washington speaks the language of innovation, and it is the one place where the technology is actually built. Its blind spot is the word it repeats most often, which is freedom, since it proclaims an open market while restricting exports, keeping its most advanced chips from crossing certain borders, even as Jensen Huang, whose company designs them, took a seat on a board in Beijing.

Beijing speaks the language of harmony and of a stability held above almost everything else, resting on a view of the world as process rather than fixed principle. The coherence between what it declares and what it funds is real, yet its blind spot has a name it will not pronounce, the individual and their rights whenever these collide with stability.

London speaks the language of existential risk, and ever since Bletchley Park it has made governments sit at one table to weigh whether the true danger is today's bias or tomorrow's catastrophe. Its limit is sociological, since it speaks for a small circle of laboratories and their founders while, on the floor below, ordinary people lose their work to systems no one has explained to them.

Singapore and the financial hubs speak the language of trustworthiness and compliance, a register that is pragmatic and free of sentiment. The trouble is that compliance is not the same as value: responsible comes to mean conforming rather than just, and ethics quietly becomes another line in the budget.

Rome speaks the language of dignity, the only dialect still using words the others have abandoned, words like the human person and the dignity of work. In May 2026 it produced Magnifica Humanitas, the first encyclical devoted entirely to artificial intelligence, and it remains the one voice at the table that funds no research and owns no laboratory. That does not make it correct, yet it does make it free, because whoever finances nothing has no budget to reconcile later with their own words.

And there is a seventh place at the table that speaks none of these languages at all. Abu Dhabi speaks capital. In July 2026 its sovereign fund MGX closed at $49 billion and took a position in nearly every American frontier lab at once, while a five gigawatt campus rises in the desert to house the models the others are still arguing about. It has no need of an ethics vocabulary, because it owns a piece of everyone else's. Yet even capital meets a wall. It can buy the models and the compute, and not the single second in which a human being decides whether to hand the choice away. That second is the one thing in the whole market that is not for sale.

For all their differences these vocabularies share one blindness, that they mistake ethics for regulation. Each answers the same question, how do we govern this, and the answer always takes the shape of another framework or another summit. None turns to the question that comes first, what happens to a human being who lives and works inside an infrastructure that has begun to think in their place.

This is the part that should matter most to anyone in public service, because it is the part no regulation reaches. Beneath every rule lies a quieter layer that lives in no document, one that shows itself in a single gesture, the instant before delegation, when a person is about to hand the machine a decision that was their own and, for the length of a breath, could still keep it. No AI Act arrives there, and neither does any safety institute or encyclical, so the only thing that arrives is habit, and just now no one is taking responsibility for how that habit is formed.

The question facing anyone who writes the rules is not which of these dialects is correct. It is whether we can admit that we have been speaking several of them all along, and that the polite fiction of one shared language is itself the first error, the one from which all the others quietly descend.

One framework, Being Human versus Human Being, archived on Zenodo, tries to hold this open. Its claim is simple, that the act of living consciously has to be examined before we touch the biological substrate the technology is busy reshaping. It does not pretend to hold the answer. It asks only that the question come first, who we are as people who live before we decide how to govern the machines that will redefine who we are as a species.

Which leaves public service a stranger task than regulation. Someone has to translate between these six languages without pretending they are one, and to keep translating right up to the last untranslatable thing, the second before a human hands over a decision that, for the length of a breath, is still theirs to keep.

Sources:
Vatican — Magnifica Humanitas
MGX $49bn, backs OpenAI/Anthropic/xAI — CNBC
US chip export controls June 2026 — CNBC
Crunchbase Q1 2026 funding
Human Being vs. Being Human