Image is AI generated
The Anthropic–Trump clash is not an isolated dispute. It is one symptom of a deeper disorder: a world in which power no longer feels obliged to conceal itself and the institutions built to constrain it are collapsing in plain sight.
We are doomed. Not by fate or lack of intelligence but because we insist on pretending that we live in a normal world.
Consider three moments from recent months. The United States government designates an Artificial Intelligence (AI) company a national security threat not for espionage or sabotage but for refusing to enable mass surveillance.
In Venezuela, the elected president is seized in a manner that bypasses every convention of international law and sovereign immunity, the legal architecture of the post-war order treated not as a constraint but as an inconvenience.
In Iran, a military campaign unfolds that reshapes a nation’s future through devastation at a scale and speed that renders the deliberative mechanisms of international response functionally irrelevant.
In each case, the forms of the normal world persist: statements are issued, hearings are convened, terms of service are invoked. But the underlying equilibrium that once made those forms meaningful has thinned to near-transparency.
This is not a coincidence. It is the texture of our moment.
The normal world and the real one
When Anthropic, the AI safety company behind the Claude models, insisted that its technology not be used for mass surveillance or fully autonomous weapons systems, the Pentagon responded by designating it a “supply-chain risk”, a label typically reserved for foreign adversaries.
President Donald Trump directed federal agencies to cease work with the company, declaring that the United States would not allow a “radical left, woke company” to dictate how its military fights and wins wars. Those running Anthropic, he suggested, “have no idea what the real world is all about”.
In response, Anthropic’s CEO Dario Amodei defended his company’s position in an interview with CBS News. A private company, he argued, has the right to decide who it sells to and under what conditions. If the government dislikes the terms, it can choose another contractor.
That, he suggested, would be the “normal way” to handle such a disagreement. He is not wrong. But he is reasoning about a “normal” world that no longer exists and in many senses never did.
In this narrow sense, Trump is correct: this is what the “real” world looks like. Power does not politely negotiate with its constraints. It overwhelms them.
In a normal world, Amodei’s logic would be unassailable. Contracts have terms. Businesses have values. Governments negotiate and power is exercised with an awareness of precedent and consequence.
But we are not in a normal world. We inhabit what I call the post-normal world, a condition in which the old conventions, courtesies and constraints that once disciplined power have been shed like a skin.
I do not use post-normal in the sense advanced by Ziauddin Sardar. I am not describing complexity, uncertainty or epistemic overload in a transitory world. I am describing something starker: a world in which the performance of norms persists but the underlying architecture that once made those forms binding has hollowed out.
Post-normality is when rules still exist but no actor powerful enough feels obliged to obey them.
This distinction is not merely rhetorical. In the post-1945 order that consolidated and institutionalised modern international norms and institutions, power operated with a degree of concealment. It clothed itself in law, procedure, diplomacy and moral justification. It maintained plausible deniability.
That era is over. In this new era, power is no longer embarrassing. It does not need to conceal itself behind technocratic euphemism or institutional choreography. It is reduced to its most elemental form: blatant, brutal and brazen.
Stephen Miller, one of the chief architects of the Trump administration’s ideological posture, captured this with disarming clarity: might is right. He did not whisper it. He stated it as fact. That bluntness is the point.
The Anthropic case, the Venezuela seizure, the conduct of the war in Iran: these are not unrelated events. They are expressions of the same underlying shift, a world in which the powerful no longer feel the need to justify themselves to the frameworks designed to contain them. The post-normal world is not an aberration. It is a revelation.
The race everyone wants to win but no one will
In a normal world, the world’s leading governments would be convening emergency summits to craft binding international treaties on AI governance, extending the frameworks developed for nuclear non-proliferation and biological research to the most consequential technology ever built. Instead, we are watching an AI arms race in which the primary metric of success is competitive supremacy and the primary casualty is restraint.
When the United States designates an ethical AI company a supply-chain risk for refusing to enable mass surveillance, it does not only punish Anthropic.
It signals to every government on earth that ethical constraints on AI are negotiable liabilities, not virtues worth protecting. Many governments will take that signal enthusiastically. They already are.
But perhaps the most disorienting inversion of the post-normal world is this: we assumed, almost instinctively, that it is governments that regulate technology. That is the architecture we inherited.
Governments set boundaries, enforce rules, hold corporations accountable to the public interest. What we have stumbled into instead is the opposite. Private corporations are now drawing red lines, insisting on terms of use, attempting, however imperfectly, to restrain the state.
The Anthropic–Pentagon dispute is not a procurement disagreement. It is a symptom of profound institutional collapse. And we should not be comforted by it, neither should we celebrate it, for this arrangement will not hold long.
The market logic is inexorable. For every Anthropic that says no, there are other contractors that will enthusiastically say yes. Corporate ethics, however sincere, are not guardrails. They are business decisions, subject to revision under sufficient pressure or sufficient money. Principled refusal is not applauded in the real world. It is punished and replaced.
When the willing sellers step forward – and they are already stepping forward – the last informal brake on the weaponisation of AI will be gone.
If governments will not regulate AI and corporations cannot be relied upon to restrain themselves or their clients, who institutes the guardrails? Who enforces them? Right now, the answer is no one. And that silence is the most dangerous thing about the world we actually inhabit.
A word for Africa
The post-normal world has given us one clarifying gift: it has stripped away the pretence of normalcy. We have never been considered equals in global affairs and, for too long, we bought into the polite fiction that we were. That fiction is gone. What remains is the bare architecture of power and we would be foolish not to read it clearly.
The seizure of a sitting head of state in open defiance of international norms. A military campaign prosecuted without effective external constraint, its intent declared without apology by Pete Hegseth, the US secretary of war, who announced that “no stupid rules of engagement” would restrain American forces in Iran. A technology race in which the rules are written by those who have already decided they will win.
For those of us in Africa and the global majority world, none of this should be surprising. This is how power has always operated when it feels sufficiently unchallenged.
What is new is not the behaviour. It is the candour.
The scramble for AI dominance is the new scramble for resources and dependency is the new colonialism. Digital infrastructure built on foreign AI systems, trained on foreign data, governed by foreign terms of service, subject to foreign geopolitical whims, is not sovereign. It is subjugation in the language of innovation. And it is being offered to us, as things always have been, with a smile.
Africa must refuse. Not with rhetoric but with investment, policy and will. The continent must build its own AI capabilities, governed on its own terms, accountable to its own people. Not because this is easy. It is not. But because in the real world, power does not yield to moral appeals. It yields to power.
The belief that international norms, ethical frameworks and appeals to justice will protect us is delusional at best and suicidal at worst.
These things matter and they will always matter but they are not sufficient and they are not guaranteed. Justice without power is a petition.
What the global majority world needs now is not only a seat at the table but the leverage to hold it. For in the real world, those who cannot shape power are shaped by it and that shaping, historically, has never been kind to us.
Akanimo Andrew Akpan holds a PhD in the ethics of AI and is co-founder of AlgoViva, an AI assurance, governance and digital rights company.